Active sources: 7. Sources represented in this queue: 6. The cron runs daily at 06:00 server time; arxiv RSS is often empty on weekends.
Linked to your results
- score 100arxiv cs.CL (NLP)arxiv:2605.24451unread
Phonetic Modeling of Dialectal Variation in Vietnamese Speech
Quan Ngoc Hoang, Long Hoang Huu Nguyen, Nghia Hieu Nguyen, Kiet Van Nguyen, Ngan Luu-Thuy Nguyen · 2026-05-26
arXiv:2605. 24451v1 Announce Type: new Abstract: Vietnamese exhibits substantial dialectal phonetic variation across Northern, Central, and Southern regions, where identical lexical items may be realized with markedly different pronunciations.
Read next because Phonetic Modeling of Dialectal Variation in Vietnamese Speech overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, text, word, line, trained, model. Source: arxiv cs.CL (NLP).
arXiv:2605.24451v1 Announce Type: new Abstract: Vietnamese exhibits substantial dialectal phonetic variation across Northern, Central, and Southern regions, where identical lexical items may be realized with markedly different pronunciations. Such variation poses challenges for automatic speech recognition (ASR) and remains difficult to model computationally due to the complex relationship between Vietnamese orthography and phonology. Existing approaches typically address dialect variability at the word level, assuming dialect-invariant mappings between spelling and pronunciation, which limits their ability to capture systematic phonetic differences. We propose a dialect-aware phonetic framework that explicitly models Vietnamese phonological structure and dialectal variation at both the vocabulary and decoding levels. The framework introduces a phonetic vocabulary that decomposes each syllable into structured phonetic components and maps them to dialect-specific IPA representations, together with a phonetic-structure decoder that jointly predicts these components. Experiments on the UIT-ViMD, a only-available dataset for multi-dialect in Vietnamese, show that the proposed approach outperforms various pre-trained baselines, \textbf{especially matches the performance of the strongest pretrained wav2ve2-base-vi-250h} across dialects while \textbf{using substantially fewer parameters and no external pretraining}. Code for experimental reproducibility will be publicly available upon the acceptance of this paper.
- score 100arxiv cs.CL (NLP)arxiv:2605.24432unread
Found in Conversation: LLMs Teach Themselves to Close the Multi-Turn Gap
Tianlang Chen, Shirley Wu, Jure Leskovec · 2026-05-26
arXiv:2605. 24432v1 Announce Type: new Abstract: Large Language Model (LLM) interactions are typically underspecified, with users clarifying all necessary details across multiple conversational turns.
Read next because Found in Conversation: LLMs Teach Themselves to Close the Multi-Turn Gap overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.24432v1 Announce Type: new Abstract: Large Language Model (LLM) interactions are typically underspecified, with users clarifying all necessary details across multiple conversational turns. Yet recent work shows that LLMs perform far worse in this multi-turn setting than in a single turn with same information being available at once, a phenomenon termed "Lost-in-Conversation." However, bridging this gap effectively remains an open problem. Here we introduce Found in Conversation (FiC), a training framework where a model teaches itself to find and recover its single-turn competence given underspecified multi-turn prompts. We develop View-Asymmetric Self-Distillation, which distills across two views of the same task information--single-turn view for the teacher, multi-turn view for the student--transferring strong single-turn behavior into weak multi-turn behavior. This requires no stronger external teacher, which is unavailable as even frontier LLMs exhibit this gap. Across model families (Llama, Qwen, Phi, and OLMo) and sizes (3B-14B), FiC recovers at least 92% of single-turn performance and reaches 100% on two Llama backbones, yielding more efficient and helpful multi-turn conversations with single-turn capabilities intact.
- score 100arxiv cs.CL (NLP)arxiv:2605.24351unread
How Much Structure Do LLMs Need? Evaluating LLMs for Bibliometric Cluster Description
Abraham Camelo-Guerrero, Jairo Diaz-Rodriguez · 2026-05-26
arXiv:2605. 24351v1 Announce Type: new Abstract: Large language models (LLMs) can support scientific literature synthesis, but remain prone to hallucinated references, uneven coverage, and weakly grounded thematic organization.
Read next because How Much Structure Do LLMs Need? Evaluating LLMs for Bibliometric Cluster Description overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: under, alignment, eval, line, rate, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.24351v1 Announce Type: new Abstract: Large language models (LLMs) can support scientific literature synthesis, but remain prone to hallucinated references, uneven coverage, and weakly grounded thematic organization. We evaluate whether bibliometric structure improves LLM-assisted synthesis by comparing six pipelines for generating cluster descriptions under different levels of evidence and structure. Using 100 published bibliometric analyses, we reconstruct Scopus corpora, extract human-written cluster descriptions, and assess outputs by human alignment, semantic coverage, clustering quality, graph quality, and reference grounding. Results show that LLMs produce descriptions semantically close to human-written ones, but are unreliable when asked to infer bibliometric structure from scratch. Performance improves when bibliometric algorithms define the clusters and the LLM interprets them. Overall, LLM-assisted bibliometric synthesis is most promising as a hybrid workflow in which algorithms provide auditable structure and LLMs generate readable descriptions.
- score 100arxiv cs.CL (NLP)arxiv:2605.24344unread
Distinguishing Right from Wrong in Debates: Attribution Analysis of Chinese Harmful Memes
Weiming Wang, Junyu Lu, Han Wang, Xiaokun Zhang, Zewen Bai, Bo Xu, Liang Yang, Hongfei Lin · 2026-05-26
arXiv:2605. 24344v1 Announce Type: new Abstract: Research on harmful meme detection has garnered significant attention, resulting in the development of numerous datasets and methods.
Read next because Distinguishing Right from Wrong in Debates: Attribution Analysis of Chinese Harmful Memes overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, under, wrong, eval, source, line, rate. Source: arxiv cs.CL (NLP).
arXiv:2605.24344v1 Announce Type: new Abstract: Research on harmful meme detection has garnered significant attention, resulting in the development of numerous datasets and methods. However, progress in detecting Chinese harmful memes lags considerably, primarily due to two challenges: first, accurately assessing a meme's harmfulness depends heavily on understanding deep cultural context; second, many memes are semantically ambiguous, making harmfulness highly subjective. To address these issues, we focus on the interpretable detection of Chinese harmful memes by constructing the first Chinese harmful meme explanation dataset, Ex-ToxiCN-MM. This dataset offers opposing interpretations, categorized as "harmful" and "non-harmful", for each meme, aiming to rigorously evaluate a model's ability to discern and comprehend ambiguous, culturally grounded content. We built a specialized knowledge base of Chinese cultural concepts and offensive vocabulary to supply models with essential prior knowledge (C-HarmKB). To address the ambiguity and lack of background knowledge in meme attribution, we have developed a comprehensive attribution analysis framework, RIKE, which includes an Attribution Knowledge Enhancement module (AKE) and a Relative Intent Reasoning module (RIR). Extensive quantitative and qualitative experiments demonstrate that our method outperforms mainstream baseline models across multiple metrics in the task of attributing harmful memes in Chinese. The code, Ex-ToxiCN-MM dataset, and Chinese Harmful Semantic Knowledge Base (C-HarmKB) involved in this study have been open-sourced at https://github.com/wimiw123/Ex-ToxiCN-MM
- score 100arxiv cs.CL (NLP)arxiv:2605.24313unread
End-to-End Intracortical Speech Decoding from Neural Activity
Owais Mujtaba Khanday, Jose A. Gonzalez-Lopez, Marc Ouellet, Alberto Galdon, Gonzalo Olivares Granados · 2026-05-26
arXiv:2605. 24313v1 Announce Type: new Abstract: Current high-performing intracortical speech neuroprostheses achieve low word error rates but typically rely on external language models during inference, increasing memory, computation, and latency.
Read next because End-to-End Intracortical Speech Decoding from Neural Activity overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, word, rect, correct, rate, without, full. Source: arxiv cs.CL (NLP).
arXiv:2605.24313v1 Announce Type: new Abstract: Current high-performing intracortical speech neuroprostheses achieve low word error rates but typically rely on external language models during inference, increasing memory, computation, and latency. In this work, we investigate whether meaningful character-level decoding is achievable without such models. We propose an end-to-end Conformer-based neural decoder trained directly on intracortical recordings from a participant with amyotrophic lateral sclerosis (ALS). Without any external language model, the system achieves a character error rate (CER) of 23.80\% on held-out validation data. Analysis shows that performance variability is driven by inter-session signal degradation, while dominant errors arise from incorrect word boundary segmentation. These results demonstrate that effective character-level decoding is possible in a fully end-to-end framework, providing a strong neural signal for downstream linguistic processing.
- score 100arxiv cs.CL (NLP)arxiv:2605.24310unread
Discovering Lexical Gaps Using Embeddings from Multilingual LLMs
Yoonwon Jung, Aaron S. Cohen, Benjamin K. Bergen · 2026-05-26
arXiv:2605. 24310v1 Announce Type: new Abstract: Lexical gaps are words that do not exist in certain languages.
Read next because Discovering Lexical Gaps Using Embeddings from Multilingual LLMs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, word, class, alignment, source, rate, compare, trained. Source: arxiv cs.CL (NLP).
arXiv:2605.24310v1 Announce Type: new Abstract: Lexical gaps are words that do not exist in certain languages. They pose challenges for building multilingual lexical resources, for machine translation, and for cross-lingual transfer. Existing lexical gap detection relies on human judgments or fixed conceptual taxonomies. We propose a data-driven framework for identifying cross-lingual lexical gaps. We extracted contextualized embeddings from Korean-English bilingual LLMs for Korean-to-English and English-to-Korean translation pairs. Combinations of LLMs, embedding types, dimensionality, and orthogonal transformations across 100 train-test splits yielded 4000 distinct embedding spaces in each source language. In each space, we computed the semantic similarity between each source word and its nearest neighbor in the target language, and compared their distribution for gap words versus non-gap words. In 94% (Korean-to-English) and 97% (English-to-Korean) of embedding spaces, gap words showed weaker cross-lingual semantic alignment than non-gap words. Logistic classifiers trained on unaligned embedding spaces can reliably separate gap words from non-gap words, achieving AUCs of 0.81 (Korean-to-English) and 0.76 (English-to-Korean) and retrieving 18/19 Korean and 26/27 English gap words. This approach provides a language-agnostic and taxonomy-free method for scalable lexical gap identification.
- score 100arxiv cs.CL (NLP)arxiv:2605.24164unread
CUNY at CLPsych 2026: A Pipeline Approach to Classification and Summarization of Mental Health Changes
Amirmohammad Ziaei Bideh, Shameed Charlomar Job, Ava Yahyapour, Alla Rozovskaya · 2026-05-26
arXiv:2605. 24164v1 Announce Type: new Abstract: We describe our submission to the CLPsych~2026 Shared Task on capturing and characterizing mental health changes through social media timeline dynamics.
Read next because CUNY at CLPsych 2026: A Pipeline Approach to Classification and Summarization of Mental Health Changes overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, class, source, line, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.24164v1 Announce Type: new Abstract: We describe our submission to the CLPsych~2026 Shared Task on capturing and characterizing mental health changes through social media timeline dynamics. To infer the dominant self-states in posts (Tasks 1.1 and 1.2), we ensemble in-context learning of three open-weight large language models using majority voting. For predicting moments of change in a timeline (Task~2), we train supervised classifiers on features derived from Task~1.1 predictions. To summarize the patterns of mood dynamics and their progression over time within a timeline (Task 3.1), we augment in-context example labels predicted by upstream systems (Tasks 1.1, 1.2, and 2), yielding performance gains over zero-shot and unaugmented in-context learning baselines. Our submission ranked first on Task~1.1, fourth on Task~1.2, fourth on Task~2, and third on Task~3.1.\footnote{The source code for the experiments is available at https://github.com/amirzia/clpsych26-cuny
- score 100arxiv cs.CL (NLP)arxiv:2605.24000unread
Toxicity in Twitch Chats: An LLM-Based Analysis Across Gaming Communities
Ronja Fuchs, Florian Rupp, Timo Bertram, Kai Eckert, Alexander Dockhorn · 2026-05-26
arXiv:2605. 24000v1 Announce Type: new Abstract: Toxicity in online gaming communities remains a persistent challenge, manifesting across genres, platforms, and player interactions.
Read next because Toxicity in Twitch Chats: An LLM-Based Analysis Across Gaming Communities overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, line, rate, trained, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.24000v1 Announce Type: new Abstract: Toxicity in online gaming communities remains a persistent challenge, manifesting across genres, platforms, and player interactions. While much research is focused on in-game toxicity, less is known about how toxic behavior varies between gaming communities on streaming platforms. To address this shortcoming, we analyze approximately 20 million chat messages from 4,452 streams, spanning seven game genres on Twitch. We categorize messages according to Twitch's toxicity taxonomy with a pre-trained Large Language Model using zero-shot classification. The taxonomy comprises four categories and eight subclasses, including harassment, discrimination, sexual content, and profanity. Our approach achieves an F1 score of 94.5% on the TextDetox dataset and demonstrates human-model agreement comparable to inter-human agreement. Our analysis reveals that 2.4% of all messages are classified as toxic, with notable differences across genres: streams of MOBA games exhibit the highest relative rate of toxicity (3.2%), and sports games show the lowest rate (2%). Furthermore, results indicate that individual games differ significantly in their toxicity distributions, even within genres, suggesting the existence of game-specific community norms and mechanics that shape toxic behavior beyond genre-level effects. These findings offer empirical insights into genre- and game-specific toxicity patterns on Twitch and can inform more targeted moderation strategies for gaming communities.
- score 100arxiv cs.CL (NLP)arxiv:2605.23969unread
SLAP: Stratified Loss-based Pruning for On-Policy Data-Efficient Instruction Tuning
Run Zou, Jianhang Ding, Yifan Ding, Wen Wu, Hao Chen, Renshu Gu · 2026-05-26
arXiv:2605. 23969v1 Announce Type: new Abstract: Instruction tuning has optimized the specialized capabilities of large language models (LLMs), but it often requires extensive datasets and prolonged training times.
Read next because SLAP: Stratified Loss-based Pruning for On-Policy Data-Efficient Instruction Tuning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, eval, compare, full, on-policy, position, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.23969v1 Announce Type: new Abstract: Instruction tuning has optimized the specialized capabilities of large language models (LLMs), but it often requires extensive datasets and prolonged training times. The challenge lies in developing specific capabilities by identifying useful data and efficiently fine-tuning. High-quality and diverse pruned data can help models achieve lossless performance at a lower cost. In this paper, we propose \textbf{SLAP}, a novel batch-aware data selection framework that evaluates the learnability of entire batch compositions rather than individual. SLAP ensures comprehensive data distribution coverage through distribution-aware stratified sampling while maximizing intra-batch diversity through relative distance optimization. By leveraging Hessian-approximated gradient information for dynamic batch selection, SLAP significantly outperforms existing state-of-the-art methods across multiple model architectures (LLaMA, ChatGLM) and diverse downstream tasks including multi-turn dialogue, multilingual translation, and question answering. Most notably, SLAP achieves superior performance with 20-40\% less training data compared to full dataset training, substantially reducing computational costs while maintaining or improving model capabilities. These results establish SLAP as a powerful approach for efficient and effective instruction tuning of large language models.
- score 100arxiv cs.CL (NLP)arxiv:2605.23924unread
Improving the Completeness and Comparability of Segment Disclosures: A Large Language Model Approach
Yue Liu, Zhiyuan Cheng, Longying Lai · 2026-05-26
arXiv:2605. 23924v1 Announce Type: new Abstract: Segment-level disclosures are a central component of financial reporting, providing insight into firms' internal organization and the allocation of economic activities across operating units.
Read next because Improving the Completeness and Comparability of Segment Disclosures: A Large Language Model Approach overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, alignment, eval, rate, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.23924v1 Announce Type: new Abstract: Segment-level disclosures are a central component of financial reporting, providing insight into firms' internal organization and the allocation of economic activities across operating units. However, segment information is often presented in both qualitative and quantitative forms, dispersed across tables and narrative sections of Form 10-K filings. Empirical research relying on structured databases faces both completeness and comparability challenges, as some firm-year observations may be missing, nested segment disclosures are not captured, and support for longitudinal and cross-firm comparability is limited. This study develops a large language model-based framework to extract segment disclosures directly from Form 10-K filings and to preserve both reportable and nested segment information. We further design a retrieval augmented system that incorporates information across multiple filings to support comparability. We use two representative settings to demonstrate its application: longitudinal analysis within a firm to interpret segment changes over time, and cross firm alignment of geographic segments across firms with different reporting structures. The results indicate that the artifact accurately extracts segment-level information and effectively addresses questions that require cross-period knowledge, demonstrating the potential of LLM-based approaches to enhance the measurement and interpretation of segment disclosures.
- score 100arxiv cs.CL (NLP)arxiv:2605.23917unread
Multi-Persona Debate System for Automated Scientific Hypothesis Generation
Jaeha Oh, Byungchan Kim, Ju Li, Yang Jeong Park, Jin-Sung Park · 2026-05-26
arXiv:2605. 23917v1 Announce Type: new Abstract: Modern scientific discovery is bottlenecked not by data scarcity, but by the inability to synthesize fragmented knowledge into actionable hypotheses.
Read next because Multi-Persona Debate System for Automated Scientific Hypothesis Generation overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, persona, rect, under, eval, line, rate, control. Source: arxiv cs.CL (NLP).
arXiv:2605.23917v1 Announce Type: new Abstract: Modern scientific discovery is bottlenecked not by data scarcity, but by the inability to synthesize fragmented knowledge into actionable hypotheses. This challenge is especially acute in battery materials research, where electrochemical performance, interfacial behavior, and manufacturing feasibility must be optimized simultaneously. Here, we present the Multi-Persona Debate System (MPDS), a literature-grounded framework for automated scientific hypothesis generation that combines literature retrieval, long-context large language model reasoning, corpus-driven persona induction, and structured multi-agent debate. MPDS constructs literature snapshots of up to 500 papers, grounds agents in role-specific evidence pools, and conducts a three-round citation-aware debate followed by moderator synthesis, enabling negotiation between personas while preserving evidence traceability. We evaluate MPDS using a temporally controlled protocol excluding direct access to target papers, including two held-out battery-materials case studies and a blinded comparison across 30 matched cases. In sodium-ion anode and all-solid-state battery cathode design tasks, MPDS recovered design logics aligned with experimentally validated solution spaces and generated more mechanistically explicit, process-aware proposals than simpler baselines. To assess the impact of personas and debate, we introduce Integrative Hypothesis Quality scoring. In ablation studies, MPDS achieved the highest mean score among five conditions, with its largest advantage in cross-perspective integration. A laboratory follow-up suggests utility as a diagnostic aid for identifying practical bottlenecks in workflows. These results indicate that structured debate over literature snapshots improves hypothesis formation under coupled engineering constraints and provides a reusable workflow for text-intensive scientific discovery.
- score 100arxiv cs.CL (NLP)arxiv:2605.23910unread
Document Classification Pattern Recognition via Information Fusion: A Systematic Review of Multimodal and Multiview Representation Approaches
Marcin Micha{\l} Miro\'nczuk · 2026-05-26
arXiv:2605. 23910v1 Announce Type: new Abstract: Information fusion is used widely to improve document classification by the integration of multiple data sources (multimodal) or representations (multiview).
Read next because Document Classification Pattern Recognition via Information Fusion: A Systematic Review of Multimodal and Multiview Representation Approaches overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, rect, under, alignment, source, line, rate. Source: arxiv cs.CL (NLP).
arXiv:2605.23910v1 Announce Type: new Abstract: Information fusion is used widely to improve document classification by the integration of multiple data sources (multimodal) or representations (multiview). However, the field lacks a unified framework, a quantitative synthesis of its effectiveness, and clear guidance for practitioners. This systematic review addresses these gaps by analysing 139 primary studies. It introduces a formal framework to structure the field, presents the results of a qualitative analysis to identify key trends, and performs a random-effects meta-analysis (to our knowledge, the first focused on document classification) to quantify performance gains. Our meta-analysis reveals that multimodal fusion improves accuracy (mean gain of +5.28 percentage points, $p=0.0016$) significantly -- the F1-score effect is directionally positive but statistically non-significant in our primary model. Multiview fusion provides consistent but modest gains for accuracy (+4.67\%), F1-score (+3.08\%), and recall (all $p<0.05$). Critically, our qualitative synthesis uncovers challenges in reproducibility in methodological rigour: only 11.8\% (multimodal) and 23.3\% (multiview) of the studies use statistical tests to validate their findings, which undermines the reliability of many of their results. This review's primary contributions are a unifying framework, the first quantitative evidence base, and data-driven guidelines. This review concludes that successful information fusion depends not on algorithmic complexity, but on the strategic alignment of the fusion method with the task context and a commitment to more rigorous validation.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24192unread
Filtered Posterior Mean Collections: A Unified Framework for Analytical Models of Diffusion Generalization
Matthew Niedoba, Berend Zwartsenberg, Frank Wood · 2026-05-26
arXiv:2605. 24192v1 Announce Type: new Abstract: The neural-network denoising functions which form the backbone of image diffusion models are remarkably consistent in their generalization behaviour across a wide variety of network architectures and training procedure hyperparameters.
Read next because Filtered Posterior Mean Collections: A Unified Framework for Analytical Models of Diffusion Generalization overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, soft, source, line, rate, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24192v1 Announce Type: new Abstract: The neural-network denoising functions which form the backbone of image diffusion models are remarkably consistent in their generalization behaviour across a wide variety of network architectures and training procedure hyperparameters. A recent line of research has sought to model the outputs of these networks by aggregating posterior weighted averages of training dataset patches. In this work, we consolidate these approaches into a unified model class which we call Filtered Posterior Mean Collections (FPMCs). We define this model class using query precision vectors, response weights, and source distributions, and illustrate that existing methods are recoverable with specific choices of these design axes. Investigating each axis in turn, we find that FPMC performance can be improved with soft relaxations of prior patch-based methods, and through augmentations of source distributions. Applying these findings to an existing FPMC, we demonstrate consistent sample improvement across three natural image datasets.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24071unread
Not All Transitions Matter: Evidence from PPO
Ajhesh Basnet · 2026-05-26
arXiv:2605. 24071v1 Announce Type: new Abstract: Training a reinforcement learning agent on-policy means collecting fresh experience at every update, and that experience comes with a hidden problem.
Read next because Not All Transitions Matter: Evidence from PPO overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, implement, alone, chain, on-policy, stage, never. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24071v1 Announce Type: new Abstract: Training a reinforcement learning agent on-policy means collecting fresh experience at every update, and that experience comes with a hidden problem. Each state in a rollout is the direct output of the previous one, causally chained together by the agent's own actions. Because of this, consecutive transitions are never truly independent. They carry overlapping information, and the gradient signal the network receives ends up far more repetitive than the batch size suggests. The same directions get reinforced over and over, the value network struggles to keep up as the policy shifts, and training becomes quietly unstable in ways that reward curves alone rarely reveal. This paper asks whether that redundancy can simply be removed. We show that randomly dropping a fixed fraction of transitions from the rollout, at the right stage so the reward signal stays intact, is enough to break the repetitive gradient structure and stabilize training. The change is minimal: one sampling step, no new components, no modification to the core algorithm, and it works with any PPO implementation. Across five environments of increasing difficulty, CartPole-v1, Acrobot-v1, LunarLander-v2, HalfCheetah-v5, and Hopper-v5, the method matches vanilla PPO on reward while producing more consistent training dynamics across KL divergence, policy entropy, and value estimates. Dropping 25% of transitions turns out to be the sweet spot: enough to disrupt the redundancy, not enough to thin the batch.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24059unread
Spectral Probe-Circuits: A Three-Step Recipe for Identifying Attention-Head Circuits in Pretrained Transformers
Yongzhong Xu · 2026-05-26
arXiv:2605. 24059v1 Announce Type: new Abstract: We present a three-step recipe for identifying attention-head circuits in pretrained transformers.
Read next because Spectral Probe-Circuits: A Three-Step Recipe for Identifying Attention-Head Circuits in Pretrained Transformers overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone". Matching terms: line, rate, recipe, control, without, trained, screen, candidate. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24059v1 Announce Type: new Abstract: We present a three-step recipe for identifying attention-head circuits in pretrained transformers. A per-head spectral signal -- the time-integrated participation ratio of each head's attention output -- ranks heads doing sustained content-dependent computation without labels or attribution gradients. A task-pattern screen filters this general indicator into a task-specific candidate circuit, and group ablation against a matched-random control completes the causal claim. We validate across an 8x parameter range (51M to 1B-active / 7B-total), two architecture families (dense, mixture-of-experts), and four pretraining pipelines. The recipe ports: a 2-6 head induction circuit is causally necessary in every model tested, with a 94-100% drop in synthetic-induction top-1 after ablation. The spectral signal is predictive without supervision: on six independent seeds of a 51M-parameter probe model, the same computation identifies the seed-specific circuit on each seed. The fraction of heads doing identifiable specialized computation is conserved at 17-19% across the Pythia family (124M to 410M), while specific induction circuits stay 3-11 heads -- sublinear in total head count. This paper is the methodology anchor of a three-paper program; companion papers extend the recipe to developmental trajectories during pretraining and to composed-task circuits where pattern selectivity decouples from task-causal structure.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24058unread
Signs Beat Floats: Low-Rank Double-Binary Adaptation for On-Device Fine-Tuning
Yoshihiko Fujisawa, Yuma Ichikawa, Yudai Fujimoto, Akira Sakai, Katsuki Fujisawa · 2026-05-26
arXiv:2605. 24058v1 Announce Type: new Abstract: On-device adaptation of large language models commonly keeps a quantized base model frozen while training and deploying a small, task-specific LoRA adapter.
Read next because Signs Beat Floats: Low-Rank Double-Binary Adaptation for On-Device Fine-Tuning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: fill, line, rate, factor, lora, language, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24058v1 Announce Type: new Abstract: On-device adaptation of large language models commonly keeps a quantized base model frozen while training and deploying a small, task-specific LoRA adapter. In the unmerged adapter-mode setting, however, the adapter is more than a compact storage module; it introduces an additional dense floating-point branch, maintains a trainable state for local updates, and acts as a unit of communication and hot-swapping.We introduce LoRDBA, a LoRA-compatible adapter that replaces both low-rank factors with binary sign carriers while representing magnitudes through lightweight, channel-wise scales, converting the dense adapter branch into two sign-accumulation matrix multiplications interleaved with channel-wise scaling. A finite-sample analysis shows that reconstruction quality is governed by the residual-to-magnitude ratio of the original LoRA factors. In adapter-mode experiments, LoRDBA outperforms low-bit baselines at matched model sizes while matching fp16 LoRA quality in selected regimes. The unmerged adapter incurs at most 8% prefill latency overhead at matched rank r=16 despite an over 10x reduction in adapter footprint, with moderate training memory overhead of approximately 1.6x that of fp16 LoRA.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24057unread
Feature Lottery? A Bifurcation Theory of Concept Emergence
Fuming Yang · 2026-05-26
arXiv:2605. 24057v1 Announce Type: new Abstract: Neural networks acquire structured representations at specific moments during training, yet identifying these transitions typically relies on retrospective, label-dependent metrics.
Read next because Feature Lottery? A Bifurcation Theory of Concept Emergence overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, under, epochs, line, compare, symmetry, language, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24057v1 Announce Type: new Abstract: Neural networks acquire structured representations at specific moments during training, yet identifying these transitions typically relies on retrospective, label-dependent metrics. We introduce a bifurcation theory of representation dynamics to detect these moments in real time. Analyzing a passive GMM probe attached to the evolving encoder, we show the onset of structure corresponds to a supercritical pitchfork bifurcation driven by the loss Hessian. The system exhibits a theoretically predictable zero-crossing ($\beta_c$) that, compared to the network's current state ($\beta$), yields a dynamic ratio $\beta(t)/\beta_c(t)$: a universal, label-free phase coordinate for representation dynamics, computable entirely from hidden states. We empirically validate four distinct transition regimes predicted by this coordinate across diverse settings: SAEs on language models (Pythia), SSL (CIFAR), and grokking (modular arithmetic). Crucially, under finite dissipation, macroscopic symmetry-breaking can lag the initial zero-crossing by orders of magnitude, which providing a rigorous dynamical account of the delayed escape observed in grokking. Microscopically, the bifurcation creates a shared unstable subspace, forcing collective symmetry breaking. We term this the "feature lottery" in SAE training: a feature's terminal interpretability becomes predictable remarkably early. By only 5% of training, early atom purity robustly predicts final convergence purity, with top-decile early atoms achieving over 12x the baseline purity at convergence. Beyond explaining concept emergence, $\beta/\beta_c$ provides a practical early-warning indicator for training health, detecting the onset of usable structure, the crystallization of feature identity, and representational collapse epochs before downstream metrics react.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24043unread
LLM-AutoSciLab: Closed-Loop Scientific Discovery via Active Experimentation with LLMs
Sanchit Kabra, Nikhil Abhyankar, Saaketh Desai, Prasad Iyer, Chandan K Reddy · 2026-05-26
arXiv:2605. 24043v1 Announce Type: new Abstract: Scientific discovery is a closed-loop process in which hypotheses guide data acquisition and observations refine the hypothesis space.
Read next because LLM-AutoSciLab: Closed-Loop Scientific Discovery via Active Experimentation with LLMs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, eval, line, trained, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24043v1 Announce Type: new Abstract: Scientific discovery is a closed-loop process in which hypotheses guide data acquisition and observations refine the hypothesis space. Yet most approaches reduce discovery to supervised learning over fixed datasets, where limited observations can support multiple plausible mechanisms that fit locally but fail to generalize. Thus, the key challenge is selecting informative observations to resolve uncertainty, shifting the focus from static inference to adaptive data acquisition. To address this, we propose LLM-AutoSciLab, a closed-loop framework that couples hypothesis generation with hypothesis-conditioned experiment selection and mechanism refinement. Rather than fitting models to passively collected data, LLM-AutoSciLab iteratively proposes plausible hypotheses, selects informative experiments to distinguish or refine them, and updates its state using the resulting evidence. To evaluate dynamic, closed-loop scientific discovery with active data acquisition, we introduce ActiveSciBench, comprising two datasets: ActiveSciBench-Chem with 57 enzyme-kinetics tasks and ActiveSciBench-GRN with 45 gene-regulatory-network tasks. These datasets model discovery as a budget-constrained process requiring adaptive experiment design, variable selection, and recovery of true mechanisms. Across NewtonBench, ActiveSciBench-Chem, and ActiveSciBench-GRN, LLM-AutoSciLab outperforms prior methods, achieving 67.6% and 35.1% symbolic accuracy on NewtonBench and ActiveSciBench-Chem, respectively, and 31.1% exact graph recovery on ActiveSciBench-GRN. Moreover, hypothesis-guided experimentation is 2-5x more sample-efficient than the strongest competing baselines. Code and data are available at: https://github.com/scientific-discovery/LLM-AutoSciLab
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24042unread
Hidden-State Privacy Has an Empty Middle
Alexander Okezue Bell · 2026-05-26
arXiv:2605. 24042v1 Announce Type: new Abstract: Of $1{,}536$ Gaussian release covariances we tested for single-layer hidden-state privacy, zero achieve both moderate utility and moderate privacy against an adaptive retrieval attacker.
Read next because Hidden-State Privacy Has an Empty Middle overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: fill, class, rect, under, width, eval, prefix, middle. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24042v1 Announce Type: new Abstract: Of $1{,}536$ Gaussian release covariances we tested for single-layer hidden-state privacy, zero achieve both moderate utility and moderate privacy against an adaptive retrieval attacker. We prove a complementary Fisher-ball lower bound: every full-rank Gaussian release at $O(1)$ Fisher utility admits a direction whose Mahalanobis signal grows linearly in hidden width, ruling out uniform Gaussian safety in the class and matching the empirical empty middle. The diagonal inverse-Fisher release $\Sigma^\star_{\mathrm{diag}}(\mathcal{K}) = (2\mathcal{K}/d)\,\mathrm{diag}(1/F_{ii})$ is the unique minimax-optimal diagonal mechanism at first-order KL budget $\mathcal{K}$ and the only release with worst-attacker top-1 $\le 0.001$ at every point of a 32 model-layer grid, but it sits on a privacy/utility edge rather than filling the middle. A generalized-eigen mechanism reaching $13\times$ Pareto reduction under Euclidean retrieval collapses to $100\%$ top-1 under the adaptive Mahalanobis attacker, and a full-trajectory sequence inverter recovers $94\%$ of clean GPT-2 prefixes but $0\%$ under $\Sigma_{\mathrm{diag}}$. A split-memory transformer trained from scratch reaches $G_{\mathrm{Mah}} \in [20, 33]$ at 90M and maintains a $6$--$24\times$ advantage over same-budget GPT baselines from 30M to 1B at a fixed-token language-modeling loss penalty; pretrained models top out at 9.3. These results reframe hidden-state release from mechanism-design within the Gaussian class to architecture or release co-design.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.23984unread
Parameter Efficient Multi-Class Intelligent Scheduling for Multimodal Online Distributed Industrial Anomaly Detection
Heqiang Wang, Weihong Yang, Zheyuan Yang, Jia Zhou, Xiaoxiong Zhong, Fangming Liu, Weizhe Zhang · 2026-05-26
arXiv:2605. 23984v1 Announce Type: new Abstract: Industrial anomaly detection has attracted significant attention as a fundamental challenge in industrial systems.
Read next because Parameter Efficient Multi-Class Intelligent Scheduling for Multimodal Online Distributed Industrial Anomaly Detection overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, under, source, line, rate, lora, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.23984v1 Announce Type: new Abstract: Industrial anomaly detection has attracted significant attention as a fundamental challenge in industrial systems. The rapid advancement of heterogeneous industrial sensors has driven industrial anomaly detection from unimodal to multimodal paradigms. However, existing methods are primarily designed for centralized and offline settings, overlooking the distributed and continuously generated data characteristic of real-world industrial environments. With the advancement of edge intelligence, modern edge devices are increasingly capable of not only data acquisition but also distributed model training, enabling collaborative intelligence across the system. Industrial anomaly detection represents a critical application in this context. Motivated by these challenges, we propose a novel framework termed Multimodal Online Distributed Industrial Anomaly Detection (MODIAD). We first present a comprehensive workflow for MODIAD and then formulate a Multi-class Intelligent Scheduling (MIS) problem to coordinate cross class model updates by balancing data sufficiency and class update frequency. To efficiently solve this problem, we design a Sequential Marginal Gain Greedy (SMG) algorithm that enables effective multi-class training under resource constraints. Furthermore, to improve the computational and communication efficiency during training, we propose an Resource Efficient Class-Wise Low Rank Adaptation (REC-LoRA) strategy, which significantly reduces system overhead while preserving detection performance. Extensive experiments on two representative multimodal industrial anomaly detection datasets, MVTec 3D-AD and Eyecandies demonstrate that the proposed approach achieves superior performance and efficiency under the MODIAD scenario.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.26000unread
Statistical Inference for Stochastic Gradient Descent Beyond Finite Variance
Jose Blanchet, Peter Glynn, Wenhao Yang · 2026-05-26
arXiv:2605. 26000v1 Announce Type: new Abstract: Stochastic gradient descent (SGD) is a foundational algorithm for large-scale statistical learning and stochastic optimization.
Read next because Statistical Inference for Stochastic Gradient Descent Beyond Finite Variance overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe". Matching terms: under, rate, implement, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.26000v1 Announce Type: new Abstract: Stochastic gradient descent (SGD) is a foundational algorithm for large-scale statistical learning and stochastic optimization. However, statistical inference based on SGD iterates remains challenging when stochastic gradients have infinite variance, as the relevant limiting distributions depend on unknown nuisance parameters. In this paper, we develop an efficient, model-agnostic methodology for constructing confidence regions from SGD trajectories that applies in both finite- and infinite-variance regimes. The procedure is based on a joint weak convergence result for the Polyak-Ruppert averaged estimator and an empirical second-moment normalizer constructed from stochastic gradients along the SGD trajectory. This joint limit yields a self-normalized statistic in which the leading tail-dependent scaling terms cancel. We then use a subsampling calibration scheme to estimate the relevant critical values, avoiding explicit estimation of tail indices, slowly varying functions, or stable-law parameters. The resulting confidence regions are straightforward to implement and are asymptotically valid under both the finite- and infinite-second-moment regimes. Simulation studies show reliable coverage in various settings, supporting the proposed method as a practical tool for uncertainty quantification in stochastic optimization.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25678unread
PAC Learning with Bandit Feedback: Sharp Sample Complexity in the Realizable Setting
Steve Hanneke, Qinglin Meng, Shay Moran, Amirreza Shaeiri · 2026-05-26
arXiv:2605. 25678v1 Announce Type: new Abstract: We study the problem of multiclass PAC learning with bandit feedback in the realizable setting.
Read next because PAC Learning with Bandit Feedback: Sharp Sample Complexity in the Realizable Setting overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: class, rect, under, correct, does, full, factor. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25678v1 Announce Type: new Abstract: We study the problem of multiclass PAC learning with bandit feedback in the realizable setting. In this framework, there is an unknown data distribution over an instance space $\mathcal{X}$ and a label space $\mathcal{Y}$, as in classical multiclass PAC learning, but the learner does not observe the labels of the i.i.d. training examples. Instead, in each round, it receives an unlabeled instance, predicts its label, and receives bandit feedback indicating only whether the prediction is correct. Despite this restriction, the goal remains the same as in classical PAC learning. We provide a general characterization of the optimal sample complexity of this problem, sharp for every concept class up to logarithmic factors. Our characterization is based on a new combinatorial dimension, termed the bandit $\mathrm{DS}$ dimension, defined via generalized combinatorial structures we call pseudo-boxes. These extend the pseudo-cubes underlying the $\mathrm{DS}$ dimension by allowing a different number of neighbors in each coordinate. In contrast to the $\mathrm{DS}$ dimension, which governs the full-information setting by counting the number of coordinates in the pseudo-cube, the bandit $\mathrm{DS}$ dimension aggregates the number of neighbors across coordinates, leading to a characterization in which the sample complexity scales with the total number of neighbors. We also propose a general learning algorithm achieving the upper bound, based on an algorithmic principle called ListCascade, which connects bandit learning to list learning and may be of independent interest.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25608unread
Learning Sparse Compositional Functions with Norm-Constrained Neural Networks
Shuo Huang, Lorenzo Fiorito, Lorenzo Rosasco, Tomaso Poggio · 2026-05-26
arXiv:2605. 25608v1 Announce Type: new Abstract: The ability of deep neural networks to learn hierarchical features is widely regarded as a key mechanism underlying their success in high-dimensional learning.
Read next because Learning Sparse Compositional Functions with Norm-Constrained Neural Networks overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, under, rate, without, trained, position, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25608v1 Announce Type: new Abstract: The ability of deep neural networks to learn hierarchical features is widely regarded as a key mechanism underlying their success in high-dimensional learning. Existing theory partially supports this view by establishing approximation rates based on parameter counts and sample complexity guarantees for compositional models without incurring the curse of dimensionality (CoD). To study overparameterized regimes, where the number of parameters exceeds the sample size, we develop a framework that measures complexity via the parameter norm. Within this approach, we establish approximation rates and excess risk bounds for learning sparse compositional functions whose compositional structure is represented by directed acyclic graphs (DAGs), using Frobenius norm-constrained deep neural networks. Our results have broad applicability since every function that is efficiently Turing computable admits sparse compositional representations. In particular, we cover a range of representative models, including multi-index models, binary tree structures, and general compositional architectures. The rates we derive show that deep networks can exploit the compositional structure of the target functions, effectively avoiding the CoD through hierarchical representations.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25592unread
Optimal Design for Multinomial Logit Model with Applications to Best Assortment Identification
Joongkyu Lee, Min-hwan Oh · 2026-05-26
arXiv:2605. 25592v1 Announce Type: new Abstract: We study optimal experimental design for multinomial logit (MNL) bandits, where an agent repeatedly selects a subset of $K$ items from a ground set of size $N$ and observes single-choice feedback.
Read next because Optimal Design for Multinomial Logit Model with Applications to Best Assortment Identification overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, line, full, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25592v1 Announce Type: new Abstract: We study optimal experimental design for multinomial logit (MNL) bandits, where an agent repeatedly selects a subset of $K$ items from a ground set of size $N$ and observes single-choice feedback. Unlike linear or generalized linear bandits, MNL bandits have a combinatorial action space, which makes classical optimal design approaches and naive optimization over all subsets computationally intractable. We propose a computationally efficient optimal design framework for MNL models that achieves both statistical efficiency and scalability through two complementary approaches: (i) an exact or certified-approximate reformulation of the design oracle as a $0$-$1$ mixed-integer linear program (MILP) with solver-certified early stopping, and (ii) a fully polynomial-time lifted design that replaces the nonlinear objective with a tractable surrogate. Using the Kiefer-Wolfowitz equivalence theorem, we establish near G-optimality guarantees and characterize the induced statistical-computational trade-offs. As an application, we develop a best assortment identification algorithm for MNL bandits with linear utilities and non-uniform revenues, and prove an instance-dependent sample complexity of $\tilde{O}\big(\frac{d \log N}{\Delta^2}\big)$, where $d$ is the feature dimension, $N$ is the number of arms, and $\Delta$ is the minimum revenue gap.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25590unread
Nonstationary Generalized Linear Bandits with Discounted Online Mirror Descent
Joongkyu Lee, Min-hwan Oh · 2026-05-26
arXiv:2605. 25590v1 Announce Type: new Abstract: We study nonstationary generalized linear bandits (GLBs), where the expected reward is modeled through a nonlinear link function with an unknown time-varying parameter.
Read next because Nonstationary Generalized Linear Bandits with Discounted Online Mirror Descent overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, line, project, length, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25590v1 Announce Type: new Abstract: We study nonstationary generalized linear bandits (GLBs), where the expected reward is modeled through a nonlinear link function with an unknown time-varying parameter. This framework encompasses a broad class of reward models, including linear, Bernoulli, and binomial rewards. Existing approaches are predominantly based on maximum-likelihood estimation (MLE), using sliding-window, restart, or discounting mechanisms to handle nonstationarity. Although these methods achieve statistically efficient regret guarantees, they generally require revisiting past observations at every round, which leads to computation and memory costs that grow with time; moreover, several of them rely on a non-convex projection step. In this paper, we propose DOMD-GLB, a new algorithm for nonstationary GLBs that utilizes discounted online mirror descent (DOMD) for parameter estimation, thereby incurring only $O(1)$ computation and memory costs per round. We prove dynamic regret bounds of order $\tilde{O} \big(c_\mu^{-1/2} d^{3/4} P_T^{1/4} T^{3/4}\big)$ in drifting environments and $\tilde{O}\big(c_\mu^{-1/3} d^{2/3} \Gamma_T^{1/3} T^{2/3}\big) $in piecewise-stationary environments, where $d$ denotes the feature dimension, $T$ the time horizon, $P_T$ the path length, $\Gamma_T$ the number of change points, and $c_\mu$ a curvature parameter associated with the link function, while substantially improving computational efficiency over prior work. To the best of our knowledge, this is the first algorithm for nonstationary GLBs with per-round computation and memory costs independent of time.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25567unread
Rao-Blackwellized Score Matching on Manifolds
Divit Rawal · 2026-05-26
arXiv:2605. 25567v1 Announce Type: new Abstract: We study denoising score matching (DSM) when the latent distribution is supported on a smooth embedded manifold $M \subset \mathbb{R}^D$.
Read next because Rao-Blackwellized Score Matching on Manifolds overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: rect, under, correct, project, factor. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25567v1 Announce Type: new Abstract: We study denoising score matching (DSM) when the latent distribution is supported on a smooth embedded manifold $M \subset \mathbb{R}^D$. Under ambient Gaussian corruption, the tangent denoising target contains a singular normal-fiber noise channel whose variance diverges as $d/\sigma^2$ as $\sigma \to 0^+$. We show that conditioning on the nearest-point projection $\pi(X)$ canonically removes this singularity: the resulting conditional expectation is the unique $L^2$-optimal Rao-Blackwellized predictor of the tangent DSM target among all estimators depending only on the projected observation $\pi(X)$. We then compute the small-noise expansion of this canonical target and show that it equals the intrinsic Riemannian score up to an explicit order-$\sigma^2$ correction that decomposes into an intrinsic Tweedie term and an extrinsic curvature term involving the Weingarten and Ricci operators. In the flat case, the construction reduces exactly to ordinary lower-dimensional Gaussian DSM, while on $S^d$ the extrinsic correction simplifies to the scalar factor $(1-d/2)\nabla_M \log q$; this extrinsic $\sigma^2$ correction cancels identically on $S^2$, though the intrinsic Tweedie term remains.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25383unread
Learning manifold diffusion semigroups from graph transition matrices
Xiuyuan Cheng, Nan Wu · 2026-05-26
arXiv:2605. 25383v1 Announce Type: new Abstract: We consider graph diffusion processes constructed from finite i.
Read next because Learning manifold diffusion semigroups from graph transition matrices overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: class, rect, under, rate, factor, test. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25383v1 Announce Type: new Abstract: We consider graph diffusion processes constructed from finite i.i.d. samples drawn from an unknown manifold embedded in ambient Euclidean space, where the graph affinity is defined by an ambient Gaussian kernel matrix. We show that the manifold heat semigroup $Q_t = e^{t\Delta}$ can be approximated directly by iterating the graph transition matrix $P$, under only low regularity assumptions on the test function $f$, including the case $f \in L^\infty$. We bound $\| P^n f - Q_t f \|$ in $\infty$-norm, with the operator application to $f$ properly defined, and we recover the classical graph-Laplacian pointwise rate $O(N^{-2/(d+6)})$ up to logarithmic factors, for diffusion times $t $ up to $O(1)$ and longer. The rate holds for in-sample error as well as out-of-sample generalization, where the estimator of $Q_t f$ at a new point is defined via kernel convolution. To handle non-uniform sampling densities on the manifold, we introduce a right-normalization of the graph transition matrix; under the assumption that the sampling density $p$ is $C^3$ and bounded away from zero, the same convergence rates hold. We numerically demonstrate the performance of the proposed estimator on simulated data.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25173unread
Nystr\"om Kernel Stein Discrepancy Tests
Florian Kalinke, Zolt\'an Szab\'o, Bharath K. Sriperumbudur · 2026-05-26
arXiv:2605. 25173v1 Announce Type: new Abstract: Kernel Stein discrepancy (KSD) is among the most popular goodness-of-fit (GoF) measures on general domains with a large number of successful deployments.
Read next because Nystr\"om Kernel Stein Discrepancy Tests overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, under, good, rate, test. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25173v1 Announce Type: new Abstract: Kernel Stein discrepancy (KSD) is among the most popular goodness-of-fit (GoF) measures on general domains with a large number of successful deployments. One of the main applications of KSD is in constructing powerful GoF tests. However, tests relying on the classical U-/V-statistic-based KSD estimators have two major drawbacks. (i) Their runtime scales quadratically in the number of samples. (ii) Their asymptotic null distribution is computationally intractable in most cases, typically handled by bootstrapping. While it is known that the Nystr\"om method permits accelerating KSD estimation with no loss of statistical accuracy under mild conditions, to the best of our knowledge, the fundamental question of its impact on bootstrap-based GoF testing is open; resolving this question is the focus of the current paper. In particular, we prove that the key properties of the quadratic-time bootstrapped KSD-based GoF test (asymptotic level and local consistency) are preserved by its Nystr\"om acceleration. We numerically demonstrate the efficiency of the accelerated KSD estimator and bootstrap in the context of GoF testing of spherical and functional data. Our numerical results show that the Nystr\"om-accelerated method performs statistically on-par with the quadratic-time approach, while requiring substantially smaller runtime.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24929unread
Estimating Mixture Distributions via Stochastic Mirror Descent
Mohammadreza Ahmadypour, Tara Javidi, Farinaz Koushanfar · 2026-05-26
arXiv:2605. 24929v1 Announce Type: new Abstract: We revisit the classical problem of estimating an unknown distribution from its samples by fitting a mixture model that minimizes cross-entropy loss.
Read next because Estimating Mixture Distributions via Stochastic Mirror Descent overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: word, class, under, rate, without, does, candidate, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24929v1 Announce Type: new Abstract: We revisit the classical problem of estimating an unknown distribution from its samples by fitting a mixture model that minimizes cross-entropy loss. Framing the task as a stochastic convex optimization problem over the space of $ M $-component mixture distributions, we propose a family of estimators derived from the stochastic mirror descent (SMD) algorithm. This optimization-based approach provides a principled and flexible framework that generalizes traditional estimators and proposes a variety of novel estimators through the choice of Bregman divergences. A key advantage of our method is that it scales efficiently with the number of candidate components $ f_i $; that is, one can employ a large set of basis distributions in the mixture model without incurring significant computational overhead. This enables richer approximations and improved estimation accuracy. Moreover, in the case of categorical distribution (discrete outcomes) our estimators do not require a strict lower bound, in other words our framework does not require the precise knowledge of the support of the distribution. We demonstrate that, under mild conditions, the proposed $ \varphi $-SMD estimators achieve near-optimal convergence rates in both Kullback-Leibler (KL) divergence and $ \ell_2 $-norm and offer practical benefits when computation is expensive. Our numerical analysis highlights improved performance guaranties over classical estimators, particularly in terms of sample efficiency and scalability.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24749unread
How Neural Reward Models Learn Features for Policy Optimization: A Single-Index Analysis
Rei Higuchi, Ryotaro Kawata, Akifumi Wachi, Shokichi Takakura, Kohei Miyaguchi, Taiji Suzuki · 2026-05-26
arXiv:2605. 24749v1 Announce Type: new Abstract: Reward modeling is not only a prediction problem: in KL-regularized policy optimization, the learned reward is exponentiated to define the deployed policy, so downstream value depends on errors in reward-tilted regions.
Read next because How Neural Reward Models Learn Features for Policy Optimization: A Single-Index Analysis overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, factor, stage, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24749v1 Announce Type: new Abstract: Reward modeling is not only a prediction problem: in KL-regularized policy optimization, the learned reward is exponentiated to define the deployed policy, so downstream value depends on errors in reward-tilted regions. We study this feedback in a Gaussian single-index model with $r^*(x) = \sigma^*(\langle \theta^*, x\rangle)$ and $x \sim N(0, I_d)$. We analyze a two-stage neural reward model that first learns the hidden direction $\theta^*$ from reward-weighted samples and then fits the readout layer by weighted ridge regression. Exponential reward weighting changes the Hermite signal available to the first layer; for any feature-learning temperature $\beta_1$ above a dimension-free $O(1)$ threshold, a constant fraction of neurons recover the hidden direction, with weak-recovery complexity governed by the generative exponent. After feature recovery, we derive tilted-policy value-gap bounds for an idealized label-weighted fit with weights $e^{y/\beta_2}$ and a more practical surrogate-weighted fit with weights $e^{r_{a_0}(x)/\beta_2}$. Keeping the $\beta_2$-dependence explicit yields an admissible set of deployment temperatures, balancing the gain from lowering $\beta_2$ against the learning cost amplified by exponential weighting; in the surrogate-weighted case, proxy-dependent factors shrink this admissible set.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24136unread
Detecting Metastable Basins in High Dimensions via Marginal Trajectory Distribution Discrimination
Taj Jones-McCormick · 2026-05-26
arXiv:2605. 24136v1 Announce Type: new Abstract: We study the problem of identifying dynamically distinct basins of attraction in high dimensional time-homogeneous Markov processes using only trajectory sampling.
Read next because Detecting Metastable Basins in High Dimensions via Marginal Trajectory Distribution Discrimination overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, under, eval, line, rate, candidate. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24136v1 Announce Type: new Abstract: We study the problem of identifying dynamically distinct basins of attraction in high dimensional time-homogeneous Markov processes using only trajectory sampling. This problem is fundamental in the analysis of metastable dynamical systems, where the process rapidly mixes within basins while transitions between basins occur rarely on the timescale of interest, or even when the state space is reducible. Existing approaches typically rely on spatial discretization or spectral analysis of estimated transition operators, which can become unreliable in high dimensional settings or when the underlying basin geometry is highly nonlinear. We propose a discriminative approach to basin identification based on marginal trajectory distribution comparison. We prove a simple risk separation result: if two initial states belong to the same basin, the Bayes-optimal classifier distinguishing their marginal trajectory distributions achieves risk close to 1/2, whereas if they lie in distinct basins, the optimal risk is close to zero. This observation reduces basin detection to a two-sample discrimination problem between marginal trajectory distributions. Motivated by this principle, we develop a neural algorithm that receives a set of candidate basin representatives and iteratively merges them by estimating classification risk with a neural network that approximates the Bayes classifier. We evaluate the method on various metastable systems. These include synthetic systems constructed by embedding low-dimensional dynamics into high dimensional noisy ambient spaces. In these settings, standard spectral and clustering-based methods often fail, while our approach accurately recovers the underlying basin structure. These results display a shortcoming of existing methods and highlight trajectory discrimination as an effective tool for identifying dynamical basins in high dimensional stochastic systems.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.25002unread
MemMark: State-Evolution Attribution Watermarking for Agent Long-Term Memory Systems
Haobo Zhang, Xutao Mao, Guangyuan Dong, Ziwei Li, Xuanbo Su, Kaijie Chen, Jing Yang, Zheng Lin · 2026-05-26
arXiv:2605. 25002v1 Announce Type: new Abstract: Memory-backed agents need provenance that can survive leaked or migrated snapshots, where logs, visible outputs, and trusted metadata may be absent.
Read next because MemMark: State-Evolution Attribution Watermarking for Agent Long-Term Memory Systems overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, wrong, line, rate, control, without, full, candidates. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.25002v1 Announce Type: new Abstract: Memory-backed agents need provenance that can survive leaked or migrated snapshots, where logs, visible outputs, and trusted metadata may be absent. We propose MemMark, a state-evolution attribution watermark that embeds an owner-controlled signal into latent memory-write decisions. At each internal LLM call, MemMark samples among admissible candidates using keyed, distribution-preserving selection, and records cryptographic commitments with signed session anchors and reveal evidence. This makes attribution depend on reproducible backend behavior rather than mutable provenance fields. Across A-Mem and Graphiti on LoCoMo, with three LLM backbones, MemMark preserves memory utility: Overall F1 retains 99.6% of the unwatermarked baseline, while BLEU-1 changes by +0.2%. It also provides usable carrier capacity, with 1.16, 1.14, and 1.26 bits of mean entropy for update-target, link-target, and semantic-realization decisions. In the snapshot-only R3 setting, MemMark recovers the full 40-bit payload from final snapshots, while wrong-key verification remains near chance. Under nine memory-lifecycle attacks, verification distinguishes tampering, evidence deletion, and partial payload recovery. These results show that robust snapshot-only attribution is feasible for long-term agent memory without surviving traces, trusted metadata, or utility-degrading.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24949unread
APT-Agent: Automated Penetration Testing using Large Language Models
William Guanting Li (University of Queensland), Alsharif Abuadbba (CSIRO Data61), Kristen Moore (CSIRO Data61), Dan Dongseong Kim (University of Queensland) · 2026-05-26
arXiv:2605. 24949v1 Announce Type: new Abstract: Penetration testing is essential to securing modern web infrastructures, yet traditional manual methods struggle to keep pace with their scale and complexity.
Read next because APT-Agent: Automated Penetration Testing using Large Language Models overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, rect, under, eval, rate, compare, full, test. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24949v1 Announce Type: new Abstract: Penetration testing is essential to securing modern web infrastructures, yet traditional manual methods struggle to keep pace with their scale and complexity. Large Language Models (LLMs) offer new opportunities for automating these tasks, but existing approaches face two persistent challenges: hallucination of technical entities and insufficient long-term contextual memory. To address these issues, we present APT-Agent, a fully automated LLM-driven penetration testing framework that systematically orchestrates reconnaissance, exploitation, and exfiltration. APT-Agent introduces a hybrid rectification module to recover hallucinated commands and a command-specific memory architecture to preserve operational context across multi-step attack sequences. We evaluate our APT-Agent on Metasploitable 2 against seven vulnerable services spanning web, database, and network protocols. APT-Agent achieves an 84.29% end-to-end exploitation success rate, compared to 48.57% (Script Kiddie) and 18.57% (PentestGPT) under matched conditions. By reducing cognitive burden and minimizing reliance on human intervention, APT-Agent represents a step toward scalable, reliable, and cognitively efficient automation for penetration testing.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24663unread
CyBOKClaw: Human-in-the-Loop CyBOK Mapping for Cybersecurity Curriculum
Yan Lin Aung, Kevin Togbe · 2026-05-26
arXiv:2605. 24663v1 Announce Type: new Abstract: This paper presents CyBOKClaw, an interpretable human-in-the-loop retrieval framework for mapping cybersecurity keywords or phrases (KWoPs) to the Cyber Security Body of Knowledge (CyBOK).
Read next because CyBOKClaw: Human-in-the-Loop CyBOK Mapping for Cybersecurity Curriculum overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: word, phrase, class, phrases, alignment, eval, rate, alone. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24663v1 Announce Type: new Abstract: This paper presents CyBOKClaw, an interpretable human-in-the-loop retrieval framework for mapping cybersecurity keywords or phrases (KWoPs) to the Cyber Security Body of Knowledge (CyBOK). Rather than treating the task as strict exact classification, the framework is designed as a top-k candidate generator for expert review. It combines query normalization, curated term expansion, concept-level boosts, topic-description enrichment, and domain-sensitive ranking rules. Because educational KWoPs are often broad, ambiguous, and only approximately aligned with CyBOK terminology, strict exact matching provides only a partial account of practical utility. We therefore evaluate the framework using both structural retrieval metrics and an expert-guided top-5 usefulness metric, ECA-5 (Exact or Closest Acceptable Match at top-5), which records whether the returned candidates contain at least one mapping that an expert would judge exact or accept as the nearest practical CyBOK placement. On the development dataset, CyBOKClaw achieves 64.73% EXA-5 (Exact Match at top-5), 84.18% structural semantic alignment, and 91.88% ECA-5; on the validation dataset, it achieves 81.19% EXA-5, 93.32% structural semantic alignment, and 98.00% ECA-5. These results show that expert-guided top-k usefulness provides a more faithful account of practical CyBOK mapping utility than exact structural matching alone, and that CyBOKClaw is effective as a CyBOK-specific expert-support retrieval system.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24632unread
Demystifying the Mythos or Disrupting Bugonomics? From Zero-Day Asymmetry to Defender Remediation Throughput
Alfredo Pesoli, Herman Errico, Lorenzo Cavallaro · 2026-05-26
arXiv:2605. 24632v1 Announce Type: new Abstract: Recent demonstrations of large language models producing candidate and confirmed vulnerabilities in production software have renewed the narrative that AI will reshape offensive and defensive security.
Read next because Demystifying the Mythos or Disrupting Bugonomics? From Zero-Day Asymmetry to Defender Remediation Throughput overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, soft, source, anth, line, chain, candidates, candidate. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24632v1 Announce Type: new Abstract: Recent demonstrations of large language models producing candidate and confirmed vulnerabilities in production software have renewed the narrative that AI will reshape offensive and defensive security. Headlines emphasize capability; they rarely interrogate costs and incentives. This paper examines LLM-driven vulnerability discovery through a bugonomics lens: the operational economics of producing, proving, prioritizing, and fixing security-relevant defects. Historically, the most visible high-end bugonomics was offense-priced because production-grade zero-days and exploit chains were expensive specialist outputs for governments, brokers, and offensive vendors. Defender-side bugonomics already existed in vulnerability research, reward programs, and vendor remediation work; LLM-assisted systems change its scale and distribution. They make candidate generation, code comprehension, harness construction, proof-of-impact drafting, and report preparation cheaper at codebase scale. Exploits and proofs of concept remain important, but in defender workflows they primarily prove impact, guide prioritization, and justify remediation. The resulting bottleneck is not only finding more bugs; it is absorbing, validating, triaging, patching, and shipping a larger stream of reports. Using public data from Anthropic's Mythos Preview and Mozilla Firefox collaborations, along with public exploit-market price anchors and vulnerability reward programs, we argue that the near-term shift is not simply more zero-days. It is a move toward broader defender remediation throughput: low-signal candidates become cheaper, evidence-rich remediation become more important, and scarce capacity shifts toward maintainer review and release work. The effect is acute in open source, where LLM-assisted discovery can increase report volume while maintainer-side validation, triage, funding, and release capacity may not scale.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24312unread
Five Queries Are Enough: Query-Efficient and Surrogate-Free Membership Inference Attacks on RAG via Entailment
Nguyen Linh Bao Nguyen, Wanlun Ma, Viet Vo, Alsharif Abuadbba, Minghong Fang, Jun Zhang, Yang Xiang · 2026-05-26
arXiv:2605. 24312v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) has become central to large language model (LLM) deployments, grounding responses in enterprise or proprietary data to reduce hallucinations.
Read next because Five Queries Are Enough: Query-Efficient and Surrogate-Free Membership Inference Attacks on RAG via Entailment overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, rate, compare, leakage, candidate, language, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24312v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) has become central to large language model (LLM) deployments, grounding responses in enterprise or proprietary data to reduce hallucinations. However, this design introduces a new privacy risk: model outputs may signal the presence of specific documents in the retrieval corpus, enabling membership inference attacks (MIAs) that leak sensitive information. Existing MIAs are feasible, but they often rely on easily detected templated queries or require many non-templated yet costly and repetitive queries, limiting practicality. We ask: Can an adversary launch a limited-budget, surrogate-free, stealthy, and defense-agnostic membership inference attack using non-templated queries? We present MEntA (Membership Entailment Attack), a query-efficient MIA that leverages natural-language entailment to maximize information gained per query. By asking low-cost, broad, information-seeking questions and measuring entailment between model responses and candidate documents, MEntA eliminates the need for costly shadow models and large query budgets. Across NFCorpus, SCIDOCS, and TREC-COVID, MEntA achieves up to 0.991 AUC with only 5 queries, outperforming prior methods by 0.20 to 0.50 AUC under equivalent conditions. It remains effective under state-of-the-art (SOTA) RAG defenses, while current detectors either miss MEntA or flag benign queries at high rates. Regarding cost, MEntA reduces total attack cost by up to 65 $\times$ lower compared to SOTA attacks under the same attack setting. Our findings expose the feasibility of realistic, low-cost privacy leakage in RAG systems and highlight the urgent need for privacy-aware retrieval and defense mechanisms.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24300unread
Enhancing Reliability in LLM-Based Secure Code Generation
Mohammed F. Kharma, Mohammad Alkhanafseh, Ahmed Sabbah, David Mohaisen · 2026-05-26
arXiv:2605. 24300v1 Announce Type: new Abstract: Large language models (LLMs) are widely used for code generation, but their security reliability remains inconsistent across languages and prompting strategies.
Read next because Enhancing Reliability in LLM-Based Secure Code Generation overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rect, correct, eval, rate, chain, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24300v1 Announce Type: new Abstract: Large language models (LLMs) are widely used for code generation, but their security reliability remains inconsistent across languages and prompting strategies. Existing prompt engineering improves functional correctness but rarely ensures consistent security outcomes. We introduce the \textit{Mitigation-Aware Chain-of-Thought (MA-CoT)} framework, which embeds task-specific CWE mitigation guidance and language-aware safeguards to reduce recurring vulnerabilities in generated code. We evaluate MA-CoT across three LLMs (gpt-5, claude-4.5, gemini-2.5), three programming languages (C, Java, Python), and four prompting strategies (Vanilla, Zero-shot, CoT, MA-CoT) on a 200-task primary dataset, with external validation on LLMSecEval. Using static analysis with expert validation, MA-CoT reduces total security findings from 92 to 39 (57.6\%) on the primary dataset and from 73 to 4 (94.5\%) on LLMSecEval. High-severity findings (Blocker + Critical) drop from 90 to 39 (56.7\%) and from 45 to 2 (95.6\%), respectively. Across both datasets, MA-CoT is the only strategy that consistently improves security reliability; Zero-shot and CoT are less reliable and may increase vulnerability, especially in C. We further introduce a strict layered attribution of vulnerability drivers (language-core vs. stack layers) and show that residual risk concentrates in hardening-oriented patterns (e.g., OS- and toolchain-dependent), motivating secure-by-construction primitives alongside prompting.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24294unread
Concept Drift Adaptation Using Self-Supervised and Reinforcement Learning In Android Malware Detection
Ahmed Sabbah, Mohammad Kharma, Mohammad Alkhanafseh, Radi Jarrar, Samer Zein, David Mohaisen · 2026-05-26
arXiv:2605. 24294v1 Announce Type: new Abstract: Android malware detectors often degrade after deployment because of concept drift, while full retraining at each maintenance step is costly.
Read next because Concept Drift Adaptation Using Self-Supervised and Reinforcement Learning In Android Malware Detection overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, class, under, eval, rate, control, full. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24294v1 Announce Type: new Abstract: Android malware detectors often degrade after deployment because of concept drift, while full retraining at each maintenance step is costly. We propose a chronological adaptive maintenance framework that models deployment-time maintenance as a sequential decision problem. The framework learns a stable latent representation through self-supervised learning during initialization, freezes the encoder, measures latent drift in the fixed representation space, and performs lightweight downstream adaptation using a trainable adapter and classification head. A proximal policy optimization controller selects low-cost maintenance actions based on the detector state, including current utility, retention on a fixed memory set, latent drift indicators, and update cost. We evaluate the framework under a causal deployment-style protocol on emulator and real Android malware datasets with static and dynamic features. Results show that the RL controller provides a strong cost-aware adaptation strategy, consistently remaining among the top-performing policies while achieving a favorable balance between temporal performance, memory retention, and maintenance cost under non-stationary deployment conditions.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24245unread
Deep-Research Agents Can Be Poisoned via User-Generated Content
Tingwei Zhang, Harold Triedman, Vitaly Shmatikov · 2026-05-26
arXiv:2605. 24245v1 Announce Type: new Abstract: Deep-research agents, i.
Read next because Deep-Research Agents Can Be Poisoned via User-Generated Content overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, eval, source, line, rate, stage. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24245v1 Announce Type: new Abstract: Deep-research agents, i.e., systems that rely on multi-agent pipelines to iteratively retrieve, synthesize, and cite Web content in order to produce structured reports, are rapidly replacing traditional search for both routine and complex information needs. These agents issue many related queries during a single research session. We show that for many common search topics, they repeatedly retrieve the same user-generated content (UGC) pages from platforms such as Reddit and Wikipedia. Next, we argue that this retrieval overlap creates a concentrated attack surface: an adversary who appends a short, crafted text to a single, frequently retrieved UGC page can cause the agent to cite attacker-chosen content and promote attacker-chosen entities across many related queries. We evaluate this attack on three representative deep-research systems (STORM, Co-STORM, and OmniThink) across multiple query clusters. We also study defenses at different stages of the pipeline, including source-level filtering and output-based detection. Our findings highlight a fundamental vulnerability in how deep-research agents retrieve and integrate web content.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24206unread
FALCON-C: Flow-based Analysis and Labeling for Connected Vehicular Network Cybersecurity
Joshua Bean, Dimitrios Michael Manias · 2026-05-26
arXiv:2605. 24206v1 Announce Type: new Abstract: Along with the recent rise in popularity of Electric Vehicles (EVs), Electric Vehicle Supply Equipment (EVSE) has emerged as a new target for cyber attacks.
Read next because FALCON-C: Flow-based Analysis and Labeling for Connected Vehicular Network Cybersecurity overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, class, rate, full, trained, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24206v1 Announce Type: new Abstract: Along with the recent rise in popularity of Electric Vehicles (EVs), Electric Vehicle Supply Equipment (EVSE) has emerged as a new target for cyber attacks. Therefore, ensuring the security and integrity of network communication between EVSE components and vehicular clients is a significant challenge that must be addressed. To this end, this paper proposes a Flow-based Analysis and Labeling for COnnected vehicular Network Cybersecurity (FALCON-C) framework. The FALCON-C framework leverages an autoencoder for anomaly detection and is trained on a small number of benign flows from the CICEVSE2024 dataset. The model's objective is to model benign flow behavior and identify malicious flows by detecting statistically different reconstruction error profiles. The results demonstrate that the model can successfully identify malicious flows, achieving 100% accuracy. Initially, some benign flows were misclassified as malicious, resulting in a suboptimal false positive rate. A thorough analysis of the autoencoder's performance and the nature of misclassified flows led to the development of a refined decision boundary, improving the framework's performance by 8.6%. FALCON-C is intended to support Security Operations Center activities by automating flow labeling, leading to the enhanced curation of reliable datasets that can be used for various activities, including threat modeling and hunting, decision auditing, and intrusion detection system refinement.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24054unread
Verifiable Secure Aggregation via Dual Servers with Linear Tags in Federated Learning
Yufei Zhou · 2026-05-26
arXiv:2605. 24054v1 Announce Type: new Abstract: Federated learning (FL) enables collaborative model training by aggregating local updates without requiring raw data sharing.
Read next because Verifiable Secure Aggregation via Dual Servers with Linear Tags in Federated Learning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, line, rate, without, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24054v1 Announce Type: new Abstract: Federated learning (FL) enables collaborative model training by aggregating local updates without requiring raw data sharing. However, prior studies have shown that servers can exploit gradient inversion to compromise user privacy or manipulate aggregation results, undermining the utility of the global model. To address these concerns, we propose a secure and verifiable aggregation scheme with lightweight cryptographic primitives for FL. Our method leverages pseudo-random functions (PRFs) and a non-colluding dual-server architecture to achieve secure aggregation with mutual server verification, while maintaining communication overhead comparable to plaintext aggregation and a constant verification tag size. Crucially, it preserves user privacy and achieves end-to-end secure aggregation with verification. Moreover, our scheme significantly reduces both user computation and verification overhead, making it suitable for FL with a large number of participants. For instance, with an input dimension of 20K, user computation time is reduced to 18 ms, approximately 7$\times$ faster than OPSA, while verification time decreases to 9.5 ms, approximately 2.4$\times$ faster than OPSA.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23987unread
Beyond Predefined Learning Objects: A Thinking-Learning Interaction Model for Up-to-Date Autonomous Robot Learning
Hong Su · 2026-05-26
arXiv:2605. 23987v1 Announce Type: new Abstract: Autonomous robots operating in open and changing environments cannot always rely on predefined inputs, outputs, and action routines.
Read next because Beyond Predefined Learning Objects: A Thinking-Learning Interaction Model for Up-to-Date Autonomous Robot Learning overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, rate, length, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23987v1 Announce Type: new Abstract: Autonomous robots operating in open and changing environments cannot always rely on predefined inputs, outputs, and action routines. Although existing learning methods enable robots to improve their performance through environmental interaction, the objects of learning are often fixed in advance, such as input features, recognition outputs, network structures, task goals, or action sequences. This limits their ability to adapt when new features, new categories, or more efficient task routines appear during long-term operation. To address this problem, this paper proposes a thinking-learning interaction model for autonomous robots. The core idea is that thinking guides learning by identifying potential changes, selecting useful evidence, organizing training materials, and planning verification actions, while learning promotes thinking by updating task knowledge, feature-selection experience, action strategies, and future reasoning processes. Based on this bidirectional mechanism, the robot can gradually move beyond predefined learning settings and adapt its recognition relations and action relations through continuous interaction with the environment. Specifically, the proposed model supports adaptive input feature discovery, output category expansion, learning model update, and action routine reconstruction. Experimental results show that the proposed model improves the final recognition accuracy from 0.419 to 0.845 in feature adaptation, achieves higher new-category formation accuracy and model-update success rate, and reduces the average action length from 13.0 to 4.0 in action routine reconstruction. In learning-enhanced thinking, the useful evidence selection rate increases from 0.272 to 0.965, indicating that learning results can effectively improve future evidence selection and reasoning.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23957unread
Low-Cost Labels, Reliable Choices: Rollout-Calibrated Hyper-Heuristics for Job Shop Scheduling
Junhao Wei, Yanxiao Li, Yifu Zhao, Zhenhong Peng, Baili Lu, Dexing Yao, Haochen Li, Qinbin He, Sio-Kei Im, Yapeng Wang, Xu Yang · 2026-05-26
arXiv:2605. 23957v1 Announce Type: new Abstract: Learning-assisted hyper-heuristics can select among dispatching rules while preserving the feasibility and interpretability of constructive Job Shop Scheduling Problem (JSSP) heuristics.
Read next because Low-Cost Labels, Reliable Choices: Rollout-Calibrated Hyper-Heuristics for Job Shop Scheduling overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, rate, candidate, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23957v1 Announce Type: new Abstract: Learning-assisted hyper-heuristics can select among dispatching rules while preserving the feasibility and interpretability of constructive Job Shop Scheduling Problem (JSSP) heuristics. Their main computational cost lies in label generation rather than model fitting, since each supervised label usually requires rolling out candidate rules from a partial schedule. We study this label-cost problem together with a reliability problem: a learned selector should not switch away from a strong default rule unless the predicted gain is credible. The proposed selector uses regret-normalized rollout labels, a contextual KNN uncertainty estimate, and a gate that acts only when the predicted improvement exceeds an uncertainty-adjusted margin. We also vary rollout depth and breadth to measure the cost-quality trade-off. On synthetic JSSP instances, the gated selector achieves the lowest mean RPD among learned selectors, remains close to the best fixed dispatching rule, and reduces Random-HH mean RPD by more than an order of magnitude.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23951unread
Methods for Formal Verification of Agent Skills: Three Layers Toward a Mechanically Checkable Capability-Containment Proof
Alfredo Metere · 2026-05-26
arXiv:2605. 23951v1 Announce Type: new Abstract: The companion paper introduced a four-level verification lattice on agent-skill manifests (unverified, declared, tested, formal) and left the top level aspirational.
Read next because Methods for Formal Verification of Agent Skills: Three Layers Toward a Mechanically Checkable Capability-Containment Proof overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rect, correct, source, assistant, project, capability, test. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23951v1 Announce Type: new Abstract: The companion paper introduced a four-level verification lattice on agent-skill manifests (unverified, declared, tested, formal) and left the top level aspirational. This paper closes that gap. We give a precise semantics for skill behaviour faithful to how a skill is consumed by an LLM-driven runtime (a deterministic script-side reachable through a non-deterministic LLM-side), state the verification problem as a capability-containment property over that semantics, and present three composable methods that together raise a skill from declared or tested to formal: (1) sound static capability-containment analysis of the script-side via abstract interpretation over a small effect lattice; (2) a refinement type system for tool-call envelopes that mechanically rejects any call whose statically-inferred capability is not in the manifest's declared set; (3) SMT-bounded model checking against the parent paper's biconditional correctness criterion, with the bound chosen so any counter-example fitting the runtime's transaction-buffer horizon is exhibited as a concrete trace. We prove the three layers composed soundly cover the parent paper's threat model modulo a single residual (the LLM's freedom to refuse to act) that the parent paper's runtime biconditional catches at session boundary. The methods reuse existing well-engineered tools (Z3, Semgrep, CodeQL, refinement-type checkers, mechanised proof assistants) rather than asking operators to build new ones, and the proof-carrying artifact extends the existing SKILL.md convention. All three methods plus the bundle producer and re-checker ship as zero-dependency JavaScript modules in the open-source enclawed framework (https://github.com/metereconsulting/enclawed; project page https://www.enclawed.com/), with 53 unit tests and an end-to-end CLI demo on a sample skill.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23943unread
Spacetime Formation under Requirements: Contextual Realization and Form-Dependent Probability
Song-Ju Kim · 2026-05-26
arXiv:2605. 23943v1 Announce Type: new Abstract: Quantum cognition often explains order effects, contextuality, and violations of the law of total probability by replacing classical probability with quantum probability on a fixed event structure.
Read next because Spacetime Formation under Requirements: Contextual Realization and Form-Dependent Probability overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, under, rate, project. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23943v1 Announce Type: new Abstract: Quantum cognition often explains order effects, contextuality, and violations of the law of total probability by replacing classical probability with quantum probability on a fixed event structure. This paper proposes a different interpretation: quantum probability is the fixed-spacetime projection of contextual spacetime formation under finite-state requirements. The framework begins not with time, space, objects, or probabilities, but with requirements such as finite representational capacity, single-state semantic stability, context-sensitive intervention, avoidance of explicit context labels, coherent world-formation, and intersubjective transformability. When these requirements cannot be realized within a single global Boolean event structure, the mismatch appears, under fixed-spacetime projection, as noncommutativity, interference, and quantum-like probability. Building on prior single-state approaches to contextuality, we reinterpret classical contextual bookkeeping cost as the fixed-spacetime shadow of contextual spacetime formation. Auxiliary memory or context labels in a classical representation correspond, in this account, to holonomy-like mismatch among locally Boolean logic-worlds. The interference term is the cross term generated when locally classical realization contributions are nontrivially glued and projected back into a fixed classical spacetime form. The result is a transcendental-operational realist account: objecthood, eventhood, probability, and spacetime are treated as forms of realization under requirements, while objectivity is defined by invariants preserved across observer- and history-dependent spacetime formations.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23942unread
A Dynamical Framework for Cognitive Processes Based on Transformations and Semantic Equivalence
Carlo Cattani, Dioneia Motta Monte-Serrat · 2026-05-26
arXiv:2605. 23942v1 Announce Type: new Abstract: This paper proposes a structural and dynamical framework for modeling cognitive processes within a cybernetic perspective.
Read next because A Dynamical Framework for Cognitive Processes Based on Transformations and Semantic Equivalence overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, class, rate, position, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23942v1 Announce Type: new Abstract: This paper proposes a structural and dynamical framework for modeling cognitive processes within a cybernetic perspective. Cognitive states are represented as elements of a state space evolving through an iterative update rule of the form \[ X_{t+1} = \pi\big(F(f(X_t))\big), \] where $f$ describes internal transformations, $F$ represents interpretative mappings, and $\pi$ enforces semantic equivalence. The model is interpreted as a feedback system integrating transformation, observation, and stabilization. A categorical formulation is introduced to capture compositional structure, while the associated dynamics are analyzed through fixed-point arguments and contraction conditions ensuring stability. To demonstrate the operational character of the framework, a computational illustration is provided, together with a qualitative analysis of the induced dynamics. A concrete linguistic application shows how context-dependent interpretation can be modeled as a trajectory toward a stable semantic class. The proposed approach connects dynamical systems, category theory, and cognitive modeling, and provides a unified representation of cognition as a feedback-driven process evolving toward invariant interpretations.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23941unread
MEMOR-E: In-Context and Fine-Tuned LLM Personalization for Alzheimer's Assistive Robotics
Maissa Abir Smaili, Eren Sadikoglu, Ransalu Senanayake · 2026-05-26
arXiv:2605. 23941v1 Announce Type: new Abstract: Alzheimer's disease is a neurodegenerative disorder marked by progressive declines in memory and language that reduce independence in daily life, motivating socially assistive robotic support.
Read next because MEMOR-E: In-Context and Fine-Tuned LLM Personalization for Alzheimer's Assistive Robotics overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, persona, eval, line, rate, control, stage, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23941v1 Announce Type: new Abstract: Alzheimer's disease is a neurodegenerative disorder marked by progressive declines in memory and language that reduce independence in daily life, motivating socially assistive robotic support. This paper presents MEMOR-E, a mobile quadruped robot with an interactive tablet interface that assists patients and caregivers through medication reminders, routine guidance, memory oriented interactions, and companionship. We evaluated the feasibility of fine tuning large language models (LLMs) to emulate stage consistent cognitive behavior and interpret responses across standard neuropsychological language tasks, using audio transcriptions from 235 Alzheimer's patients and synthetically generated healthy controls. We also report findings on using in context learning (ICL) in LLMs, where a second LLM produced domain and severity level cognitive error summaries. Our results show that MEMOR-E can generate stage aware, non diagnostic cognitive summaries that support personalized assistive interactions, while explainable AI mechanisms translate model outputs into transparent, human readable evidence to enable caregiver oversight and trustworthy human robot interaction.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23938unread
Authority Inversion in LLM-Mediated Ubiquitous Systems: When Models Trust Users Over Sensors
Long Zhang, Zi-bo Qin, Wei-neng Chen · 2026-05-26
arXiv:2605. 23938v1 Announce Type: new Abstract: Large language models (LLMs) increasingly fuse heterogeneous inputs in ubiquitous systems.
Read next because Authority Inversion in LLM-Mediated Ubiquitous Systems: When Models Trust Users Over Sensors overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, rect, alignment, correct, eval, line, rate, control. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23938v1 Announce Type: new Abstract: Large language models (LLMs) increasingly fuse heterogeneous inputs in ubiquitous systems. Yet, how LLMs implicitly allocate authority when sensor measurements and user claims conflict remains unexamined, raising critical reliability concerns for deployments where physical sensing must retain priority. Unlike explicit traditional fusion, LLMs bury authority allocation within learned representations. We discover this allocation is severely format-dependent: numerical sensor data fails to integrate into answer-relevant model directions, allowing natural-language claims to dominate the final decision, a phenomenon we term \textbf{Authority Inversion}.To diagnose and mitigate this, we develop a geometric framework of context integration, introduce two computable audit metrics, specifically the Context Integration Ratio (CIR) and Authority Alignment Index (AAI), and propose Geometric Authority Calibration (GAC), an inference-time layer-level intervention to suppress misplaced user authority. Evaluating four models (4B to 35B parameters, three architectures) across four datasets totaling 576 conflict instances reveals extreme inversion: on numerical tasks, models exhibit near-zero sensor trust (AAI = -0.805, Cohen's d = -2.14), unaffected by model capacity. Validating our geometric framework, theory-guided causal injection flips 80.2\% of incorrect decisions (vs. <0.4\% for random controls). Practically, GAC improves HAR accuracy from 0 -- 1.6\% to 21.9 -- 27.5\%, outperforming prompting baselines. Ultimately, authority allocation in LLM-mediated systems must be explicitly audited and application-specifically configured rather than left implicit.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23937unread
BoxLitE: A Faithful Knowledge Base Embedding Based on Convex Optimization
Bruno F. Louren\c{c}o, Hesham Morgan, Ana Ozaki, Aleksandar Pavlovi\'c, Emanuel Sallinger · 2026-05-26
arXiv:2605. 23937v1 Announce Type: new Abstract: Knowledge base (KB) embeddings aim at combining the capability of classical knowledge graph embeddings to generalize the information present in facts, the ABox, with conceptual knowledge represented in an ontology language, the TBox.
Read next because BoxLitE: A Faithful Knowledge Base Embedding Based on Convex Optimization overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, capability, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23937v1 Announce Type: new Abstract: Knowledge base (KB) embeddings aim at combining the capability of classical knowledge graph embeddings to generalize the information present in facts, the ABox, with conceptual knowledge represented in an ontology language, the TBox. Several authors have recently explored the idea of mapping concepts to convex regions in a vector space. This is useful to represent hierarchies, typically present in TBoxes, since more general concepts can be mapped to larger regions, containing those regions associated with more specific concepts. However, the power of convexity is rarely leveraged during the actual learning tasks. Here, we introduce BoxLitE, a KB embedding model for DL-Lite$^{\mathcal{H}}$ that allows for convex optimization. We show that for any satisfiable DL-Lite$^{\mathcal{H}}$ KB, there is a BoxLitE embedding that is a weakly faithful model. As a proof of concept, we show how to formulate the KB embedding task as a convex optimization problem and how to obtain embeddings with such desirable faithfulness properties.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23936unread
Fuzzy, Neutrosophic, and Uncertain Graph Theory: Properties and Applications
Takaaki Fujita, Florentin Smarandache · 2026-05-26
arXiv:2605. 23936v1 Announce Type: new Abstract: This book presents a comprehensive and systematic survey of graph theory under uncertainty, with particular emphasis on the unifying role of the uncertain graph framework.
Read next because Fuzzy, Neutrosophic, and Uncertain Graph Theory: Properties and Applications overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, under, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23936v1 Announce Type: new Abstract: This book presents a comprehensive and systematic survey of graph theory under uncertainty, with particular emphasis on the unifying role of the uncertain graph framework. It reviews fundamental concepts, structural properties, graph classes, and graph parameters within fuzzy, neutrosophic, and related models, while also introducing a wide range of extensions such as uncertain digraphs, hypergraphs, superhypergraphs, and dynamic graphs. In addition to theoretical developments, the book explores practical applications, including uncertain molecular graphs, decision-making systems, graph neural networks, knowledge graphs, and cognitive maps. By organizing diverse uncertainty-aware graph models within a common perspective, this work provides a coherent framework for understanding their relationships, capabilities, and applications in complex systems.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23935unread
Operationalizing Reconstructive Authority: Runtime Construction, Dependency Resolution, and Execution Gating in Autonomous Agent Systems
Marcelo Fernandez - TraslaIA · 2026-05-26
arXiv:2605. 23935v1 Announce Type: new Abstract: Autonomous agent systems fail not only due to incorrect decisions, but due to executing decisions whose authority no longer holds at runtime.
Read next because Operationalizing Reconstructive Authority: Runtime Construction, Dependency Resolution, and Execution Gating in Autonomous Agent Systems overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: rect, correct, eval, rate, control, without, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23935v1 Announce Type: new Abstract: Autonomous agent systems fail not only due to incorrect decisions, but due to executing decisions whose authority no longer holds at runtime. Prior work defined Reconstructive Authority (RAM) as a condition for valid execution: actions are permitted only if authority can be constructed from current state. This paper addresses enforcement at runtime: how to enforce this condition in a running system. We introduce a runtime execution model in which authority is evaluated at action time and execution is conditioned on its constructibility. This extends the execution state space beyond admit/deny with a third state, halt, representing cases where authority is undefined due to incomplete or uncertain observability. We define a concrete execution protocol including dynamic dependency resolution, authority reconstruction, and explicit decision semantics. We further introduce a Recovery Loop that integrates drift detection (IML) with execution control (ACP), allowing the system to suspend execution, acquire missing information, and re-attempt authority reconstruction. We show that this model guarantees safety -- no action is executed without constructible authority -- and conditional liveness: execution resumes when authority-defining variables become observable. This work operationalizes reconstructive authority as a runtime enforcement mechanism, providing the execution semantics required to apply RAM in real systems.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23934unread
Practical Quantum CIM Empowerment via All-Domestic-Core Agentic Large Model
Wang Rui, Lu Diannan · 2026-05-26
arXiv:2605. 23934v1 Announce Type: new Abstract: Quantum computing devices are recognized as powerful tools for solving NP-complete problems.
Read next because Practical Quantum CIM Empowerment via All-Domestic-Core Agentic Large Model overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?". Matching terms: rate, implement, full, chain, stage, capability, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23934v1 Announce Type: new Abstract: Quantum computing devices are recognized as powerful tools for solving NP-complete problems. However, the intricacy of their modeling presents notable barriers for non-specialists, while the tedious iteration of constraint weights and modeling methodologies also consumes substantial effort on the part of experts. To address these challenges, this study integrates a femtosecond laser-pumped Coherent Ising Machine (CIM) with an LLM-driven agentic system by leveraging the LangGraph and LangChain frameworks. Comprehensive investigations demonstrate that large language models (LLMs) can effectively perform such tasks in modeling as QUBO/Ising model calibration, constraint weight decision iteration and rapid validation of literature-reported schemes. Notably, all these tasks can be fully implemented based on domestic large models, combined with domestically developed CIM hardware, we truly achieve the practical empowerment of quantum CIM that fully relies on all-domestic agentic large models and hardware. This work successfully realizes robust technological integration, laying a solid foundation for subsequent research. Nevertheless, it also identifies the persisting challenges in the two cutting-edge fields of large models and quantum computing at the current stage. Encouragingly, we unexpectedly discover a promising new paradigm where accumulated knowledge from agent-assisted quantum computing iterations reciprocally enhances the agent's own problem-solving capability, thereby addressing these challenges.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23930unread
Quantum Frog: Emergent Cooperation and Difficulty Scaling in a Quantized-Time Cooperative Game
Saad Mankarious · 2026-05-26
arXiv:2605. 23930v1 Announce Type: new Abstract: We introduce \emph{Quantum Frog}, a two-player cooperative game built on a novel \emph{quantized-time} mechanic in which the environment advances only when a player acts.
Read next because Quantum Frog: Emergent Cooperation and Difficulty Scaling in a Quantized-Time Cooperative Game overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, latin, rect, eval, rate, alone, does, length. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23930v1 Announce Type: new Abstract: We introduce \emph{Quantum Frog}, a two-player cooperative game built on a novel \emph{quantized-time} mechanic in which the environment advances only when a player acts. Inspired by the classic arcade game Frogger, Quantum Frog requires two frogs to cross an 8$\times$8 grid of traffic and reach the far side together. We use reinforcement learning (RL) as an analytical lens to answer four design questions: (1) how does game difficulty scale with traffic density, (2) what is the optimal single-agent policy and why, (3) how large is the cooperation gap between independent and cooperative two-agent play, and (4) what joint strategy emerges when agents are incentivised to cooperate? We train agents through five escalating stages, Tabular Q-Learning, Deep Q-Network (\DQN), Independent \DQN~(\IDQN), and Multi-Agent Proximal Policy Optimisation (\MAPPO\ with a centralised critic), evaluating each against traffic densities of one to six cars. Our key findings are: (i) the quantized-time mechanic makes a \emph{rush strategy} (moving directly upward at every step) universally optimal, as time exposure to traffic is minimised; (ii) adding an uncoordinated second player is harder than sextupling the traffic for a single expert player; (iii) cooperative training recovers +32--34 percentage points of joint success rate relative to independent agents and reduces episode length from $\sim$90 to $\sim$6 steps; and (iv) the emergent cooperative strategy is synchronised rushing, not complex positional coordination, illustrating that shared incentives alone suffice to align agents in time-critical cooperative tasks. These findings provide concrete, empirically grounded guidance for the commercial design of Quantum Frog and offer broader insights into the role of environment mechanics in shaping multi-agent learning dynamics.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23929unread
Toward Reliable Design of LLM-Enabled Agentic Workflows: Optimizing Latency-Reliability-Cost Tradeoffs
Ya-Ting Yang, Quanyan Zhu · 2026-05-26
arXiv:2605. 23929v1 Announce Type: new Abstract: Modern AI systems increasingly rely on workflows composed of multiple interacting agents, some powered by large language models (LLMs) and others by conventional computational modules.
Read next because Toward Reliable Design of LLM-Enabled Agentic Workflows: Optimizing Latency-Reliability-Cost Tradeoffs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: fill, under, token, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23929v1 Announce Type: new Abstract: Modern AI systems increasingly rely on workflows composed of multiple interacting agents, some powered by large language models (LLMs) and others by conventional computational modules. This paper analyzes the fundamental tradeoffs between latency, reliability, and cost in LLM-enabled agentic workflows. We introduce performance models for both LLM and non-LLM agents that capture the relationship between computational effort and output quality, incorporating the impact of reasoning and output tokens for LLM agents using a parametric exponential reliability function. Then, we study the design of sequential workflows under latency and cost constraints. Main results include a water-filling token allocation policy and characterizations of optimal workflow reliability in terms of shadow prices.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23928unread
Context: Proactive Goal-Directed Intelligence via Composable Sandboxed Programs, Declarative Wiring, and Structured Interaction
Gregory Magarshak · 2026-05-26
arXiv:2605. 23928v1 Announce Type: new Abstract: We present Context, the intelligence layer of the Magarshak Architecture, which replaces reactive query-response chatbots with proactive goal-directed agents that advance shared tasks without waiting for user prompts.
Read next because Context: Proactive Goal-Directed Intelligence via Composable Sandboxed Programs, Declarative Wiring, and Structured Interaction overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, rect, correct, source, rate, implement, without, emit. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23928v1 Announce Type: new Abstract: We present Context, the intelligence layer of the Magarshak Architecture, which replaces reactive query-response chatbots with proactive goal-directed agents that advance shared tasks without waiting for user prompts. The architecture rests on three mutually reinforcing mechanisms. Write-time context assembly precomputes enriched typed attributes via Groker agents, assembling interaction context as a deterministic pure function of graph state; context blocks are byte-identical across turns between semantic changes, enabling near-100% KV-cache reuse. Composable sandboxed wisdom programs form a governed library of LM-generated imperative programs declaratively wired to goal types via typed stream relations, composed via phase ordering, and executed at interaction time without further LM calls. Proactive goal stream state machines drive conversations toward terminal states by inspecting graph state and emitting structured interaction content (option arrays, governance affordances, clarification prompts) without awaiting user input. We prove six formal results: the Context Stability Theorem, bounding per-turn LM cost as a function of semantic change rate; a Program Composition Correctness Theorem; a Declarative Wiring Soundness Theorem; the Proactive Dominance Theorem, proving proactive agents weakly dominate reactive agents on expected turns-to-terminal-state; Coordination Overhead Elimination and Quality Preservation, establishing Pareto improvements in multi-participant goal chats; and a Cross-Platform Vote Consistency Theorem. Implemented in the open-source Qbix / Safebox / Safebots stack.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23909unread
Confidence Calibration in Large Language Models
Noam Michael, Daniel BenShushan, Jacob Bien, Don A. Moore · 2026-05-26
arXiv:2605. 23909v1 Announce Type: new Abstract: We investigate the calibration of large language models' (LLMs') confidence across diverse tasks.
Read next because Confidence Calibration in Large Language Models overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, rate, test, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23909v1 Announce Type: new Abstract: We investigate the calibration of large language models' (LLMs') confidence across diverse tasks. The results of our preregistered study show that the current crop of LLMs are, like people, too sure they are right: confidence exceeds accuracy, on average. Importantly, however, this tendency is moderated by a powerful hard-easy effect, wherein overconfidence is greatest on difficult tests; by contrast, easy tests actually show substantial underconfidence. We develop LifeEval, a test for evaluating model calibration across levels of difficulty.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23908unread
In Search of the Ingredients of Open-Endedness: Replicating Picbreeder with Large Vision-Language Models
Sam Earle, Kay Arulkumaran, Andrew Dai, Akarsh Kumar, Julian Togelius, Sebastian Risi · 2026-05-26
arXiv:2605. 23908v1 Announce Type: new Abstract: We are in the midst of large-scale industrial and academic efforts to automate the processes of scientific, technological and creative production through AI-driven assistants.
Read next because In Search of the Ingredients of Open-Endedness: Replicating Picbreeder with Large Vision-Language Models overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, assistant, line, rate, factor, lora, language, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23908v1 Announce Type: new Abstract: We are in the midst of large-scale industrial and academic efforts to automate the processes of scientific, technological and creative production through AI-driven assistants. Historically, a fundamental property of these processes in their human form has been their open-endedness: their capacity for generating a seemingly endless supply of novel and meaningful new forms. Do artificial agents have any capacity for such fruitful unguided discovery? To answer this question, we turn to Picbreeder, the canonical exemplar of human-driven open-ended search, in which users collaboratively generated a diverse library of images through interactive evolution of small neural networks. We replicate Picbreeder, replacing human users with frontier Vision Language Models (VLMs). We observe clear qualitative differences between the output of our system and the historical human baseline, and attempt to characterize them using metrics of phylogenetic complexity and visual and semantic salience and novelty. In an effort to identify some of the causal factors contributing these differences, we study the addition of exploratory noise to the agents' selection process, of behavioral diversity between agents, and of narrative momentum in the form of memory of past actions. We make our code available at https://github.com/smearle/picbreeder-vlm.
- score 90arxiv stat.ML (Machine Learning)arxiv:2605.24673unread
Affinity Graph Connectivity in Convex Clustering
Sam Rosen, Jason Xu · 2026-05-26
arXiv:2605. 24673v1 Announce Type: new Abstract: We generalize finite-sample bounds for convex clustering to the setting where affinity weights appearing in the objective correspond to a general connected graph.
Read next because Affinity Graph Connectivity in Convex Clustering overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: under, rate, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24673v1 Announce Type: new Abstract: We generalize finite-sample bounds for convex clustering to the setting where affinity weights appearing in the objective correspond to a general connected graph. These bounds and their analysis lead to a better understanding of clustering behavior under various implied connectivity structures behind the data and to new rates of convergence for centroid recovery. The new theoretical framework is based on random walks, which allow application of concentration inequalities related to random graph models, and formalizes the relationship between the clustering performance and the connectivity of the graph structures. Through the form of the bound and empirical results, we argue proper tuning of hyperparameters to convex clustering problems should also include tuning of input affinity weights.
- score 90arxiv stat.ML (Machine Learning)arxiv:2605.24072unread
Optimal Non-Asymptotic Edgeworth Expansions for Multivariate Neural Network Outputs
Lucia Celli · 2026-05-26
arXiv:2605. 24072v1 Announce Type: new Abstract: Finite-width fully connected neural networks with Gaussian-initialized weights deviate from their infinite-width Gaussian limit, exhibiting non-vanishing higher-order cumulants.
Read next because Optimal Non-Asymptotic Edgeworth Expansions for Multivariate Neural Network Outputs overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: width, eval, full. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24072v1 Announce Type: new Abstract: Finite-width fully connected neural networks with Gaussian-initialized weights deviate from their infinite-width Gaussian limit, exhibiting non-vanishing higher-order cumulants. We approximate these deviations, for a neural network evaluated in a finite number of inputs, using multidimensional Edgeworth expansions of arbitrary order $4m-1$, with $m\in\mathbb{N}$. Assuming that the corresponding Gaussian limit has an invertible covariance matrix and that the activation function is polynomially bounded, we establish a bound of order $n^{-m}$ on the total variation distance between the law of the true network output and its Edgeworth approximation, with matching lower bounds. As an application, we quantify the error in Bayesian posterior distributions when the prior is replaced by its Edgeworth expansion. Our results are more general and also apply to sequences of conditionally Gaussian vectors converging to a Gaussian vector with invertible covariance.
- score 90arxiv cs.AI (Artificial Intelligence)arxiv:2605.23944unread
Right-Sizing Communication and Recommendation Set Size in AI-Assisted Search
Jing Dong, Prakirt Raj Jhunjhunwala, Yash Kanoria · 2026-05-26
arXiv:2605. 23944v1 Announce Type: new Abstract: We model the interaction between a user and an AI driven recommendation system.
Read next because Right-Sizing Communication and Recommendation Set Size in AI-Assisted Search overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, assistant, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23944v1 Announce Type: new Abstract: We model the interaction between a user and an AI driven recommendation system. The user initiates the process by conveying preference information through a costly and noisy message. The AI assistant, acting as a Bayesian agent, interprets the user's message to form a posterior belief about their true preferences and make product recommendations. In particular, it determines how many recommendations to present so as to maximize the user's expected utility from their final choice, while accounting for the search cost induced by the size of the recommendation set. We use mutual information based cost functions to model the two distinct costs incurred by the user during the interaction: (i) a communication cost, which increases with the precision of their preference message, and (ii) a search cost, which increases with the size of the recommendation set provided by the AI assistant. We study products and preferences which live in d dimensional space, and ask how the user's expected payoff can be maximized. For large d, we characterize how optimal message precision and recommendation set size depend on the cost parameters, under two distinct distributions from which recommendations can be sampled from the product universe: (i) Bayes' posterior belief, and (ii) an optimized tilted distribution. Under the posterior sampling scheme (i), we identify a hybrid regime, in which an efficient interaction policy requires jointly optimizing the amount of information (in bits) conveyed by the user and the number of recommendations provided by the AI assistant. In the tilted sampling scheme (ii), our results show that the optimal interaction policy uses only one of communication and search, favoring whichever of them is less costly.
- score 82arxiv stat.ML (Machine Learning)arxiv:2605.25114unread
Counterfactually Safe Reinforcement Learning
Jingyi Li, Peng Wu, Chengchun Shi · 2026-05-26
arXiv:2605. 25114v1 Announce Type: new Abstract: Reinforcement learning algorithms are generally designed to maximize the expected return across a population.
Read next because Counterfactually Safe Reinforcement Learning overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: line, rate, control, stage. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25114v1 Announce Type: new Abstract: Reinforcement learning algorithms are generally designed to maximize the expected return across a population. However, a policy that is optimal on average may be suboptimal for certain individuals, leading to potential safety concerns. To address this, we first formalize the notion of individual harm from a counterfactual perspective and define harm as the event in which a chosen action results in a strictly worse outcome than a baseline alternative. We then propose a general two-stage procedure for learning policies that maximize the expected return while accounting for individual harm. We further establish the finite-sample properties of the learned policy, derive an upper bound on its sub-optimality gap, and show that the harm rate remains well-controlled. Numerical experiments on both simulated and real-world datasets demonstrate the effectiveness of the proposed approach.
- score 78arxiv stat.ML (Machine Learning)arxiv:2605.25526unread
From DPPs to $k$-DPPs: identifiability analysis via spectral decomposition
Hideitsu Hino, Keisuke Yano · 2026-05-26
arXiv:2605. 25526v1 Announce Type: new Abstract: We study the geometry of determinantal point processes (DPPs) through the spectral decomposition $L=U\Lambda U^{\top}$.
Read next because From DPPs to $k$-DPPs: identifiability analysis via spectral decomposition overlaps with experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: full, position, never. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25526v1 Announce Type: new Abstract: We study the geometry of determinantal point processes (DPPs) through the spectral decomposition $L=U\Lambda U^{\top}$. The spectrum $\Lambda$ governs the cardinality distribution via elementary symmetric polynomials, while the eigenspace orientation $U$ governs the conditional law within each fixed-cardinality stratum. Conditioning on cardinality $k$ yields the $k$-DPP, for which the identifiability structure changes fundamentally: the spectral parameter becomes identifiable only up to a common scale, and the eigenspace rotation parameter is identifiable only through squared minors of the eigenvector matrix. We characterize the identifiability gap precisely, via three explicit invariances (scale, sign similarity, and eigenspace rotation) and a dimension-counting theorem showing the existence of additional continuous non-identifiability whenever $\binom{N}{k}<N(N+1)/2$. In contrast, for the full DPP the non-identifiability comes only from the discrete sign similarity.
- score 78arxiv stat.ML (Machine Learning)arxiv:2605.25460unread
Mean-Shift PCA by Knockoff Mean
Mengda Li, Zeng Li, Jianfeng Yao · 2026-05-26
arXiv:2605. 25460v1 Announce Type: new Abstract: Removing noise is difficult, but adding noise is easy.
Read next because Mean-Shift PCA by Knockoff Mean overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: rate, stage, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25460v1 Announce Type: new Abstract: Removing noise is difficult, but adding noise is easy. In this work, we show how to eliminate mean-shift noisy components from PCA by deliberately introducing knockoff mean-shift perturbation. Standard PCA is highly sensitive to shifts in the sample mean: a small fraction of samples from a shifted distribution can cause large deviations in the leading principal components. In high-dimensional regimes, existing Robust PCA approaches cannot handle the mean-shift contamination structure inherent in the mixture model. Using tools from Random Matrix Theory, we prove that the mean-shift spikes are spectrally separable from the stable eigenvalues of the original covariance. Furthermore, the original eigenspace remains asymptotically invariant to the contamination, independent of the mixture weight. Exploiting this spectral stability, we propose a simple, two-stage PCA algorithm by adding knockoff mean that identifies and removes the mean-shift component using only standard PCA operations.
- score 74arxiv cs.LG (Machine Learning)arxiv:2605.24084unread
Verified SHAP: Provable Bounds for Exact Shapley Values of Neural Networks
David Boetius, Shahaf Bassan, Guy Katz, Stefan Leue, Tobias Sutter · 2026-05-26
arXiv:2605. 24084v1 Announce Type: new Abstract: Shapley additive explanations (SHAP) are widely recognised as computationally intractable for neural networks, since they induce an exponential search space over the input features.
Read next because Verified SHAP: Provable Bounds for Exact Shapley Values of Neural Networks overlaps with clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: eval, rate. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24084v1 Announce Type: new Abstract: Shapley additive explanations (SHAP) are widely recognised as computationally intractable for neural networks, since they induce an exponential search space over the input features. In this work, we take a first step towards scaling exact SHAP computation to larger search spaces by introducing an algorithm that leverages recent advances in neural network verification to compute arbitrarily tight exact lower and upper bounds on SHAP values for neural networks, ultimately recovering the exact SHAP values. We demonstrate that our approach scales to orders of magnitude larger search spaces than state-of-the-art exact methods. This provides an important first step towards exact SHAP computation and establishes a principled cornerstone for evaluating statistical approximation methods on larger search spaces.
- score 58arxiv cs.CR (Cryptography and Security)arxiv:2605.24559unread
Analyzing Concentration, Temporal Routines and Targeting in Public Ransomware Leak Site Data
Lea M\"uller (Fraunhofer Institute for Secure Information Technology, National Research Center for Applied Cybersecurity ATHENE), York Yannikos (Fraunhofer Institute for Secure Information Technology, National Research Center for Applied Cybersecurity ATHENE) · 2026-05-26
arXiv:2605. 24559v1 Announce Type: new Abstract: Ransomware has grown to become one of the most damaging types of cybercrime, affecting private and public organizations in any sector.
Read next because Analyzing Concentration, Temporal Routines and Targeting in Public Ransomware Leak Site Data overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: rate. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24559v1 Announce Type: new Abstract: Ransomware has grown to become one of the most damaging types of cybercrime, affecting private and public organizations in any sector. While early types of ransomware targeted many victims via automated attacks, ransomware groups have started to specifically target organizations and companies in the expectation of receiving larger ransoms. To increase the pressure on victims, most groups host so-called data leak sites, where information about their victims is made public. The shift towards 'human-operated' ransomware together with easily accessible behavioral traces available from data leak sites makes research investigating operational regularities of ransomware groups of interest. Using leak site posts as behavioral traces of ransomware groups, we created a dataset consisting of over 27,000 posts from 325 groups. Based on this dataset, we analyzed victim concentration, temporal routines and targeting regularities. Our findings suggest that groups do not behave entirely random. Instead, the observable traces found on leak sites show concentration of activity, temporal routines and selective patterns.
Threats and caveats
- score 100arxiv cs.CL (NLP)arxiv:2605.24452unread
Temporal Concept Drift in Legal Judgment Prediction: Neural Baselines Across Three Epochs of Ukrainian Court Decisions
Volodymyr Ovcharov · 2026-05-26
arXiv:2605. 24452v1 Announce Type: new Abstract: Legal NLP benchmarks evaluate models on randomly split data, implicitly assuming that legal language is stationary.
Read next because Temporal Concept Drift in Legal Judgment Prediction: Neural Baselines Across Three Epochs of Ukrainian Court Decisions overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, eval, epochs, line, does, full, trained, test. Source: arxiv cs.CL (NLP).
arXiv:2605.24452v1 Announce Type: new Abstract: Legal NLP benchmarks evaluate models on randomly split data, implicitly assuming that legal language is stationary. We test this assumption by fine-tuning four transformer encoders -- XLM-RoBERTa (base and large) and their legal-domain variants -- on Ukrainian court decisions from three temporal epochs defined by geopolitical disruptions: pre-war (2008-2013), hybrid war (2014-2021), and full-scale invasion (2022-2026). Each model is trained on one epoch and evaluated on all three, producing a 3x3 cross-temporal generalization matrix. Four findings emerge. (1) Forward degradation is severe: models trained on pre-war data lose up to 27.2 percentage points of macro-F1 when applied to full-scale invasion era decisions. (2) The degradation is asymmetric: backward transfer (full-scale to pre-war) is substantially more robust than forward transfer, consistent with the hypothesis that legal language is additive. (3) Legal-domain pretraining (Legal-XLM-R) does not improve absolute performance but reduces forward degradation magnitude and asymmetry. (4) Chronological continual learning eliminates catastrophic forgetting for general XLM-R: pre-war knowledge is fully retained (+1.8 to +6.2 pp) while full-scale performance gains +16.5 to +19.0 pp; reverse-chronological training causes severe forgetting. Cross-jurisdictional pretraining on Swiss Judgment Prediction data improves absolute performance but does not reduce temporal degradation magnitude, confirming that temporal drift is an intrinsic property of legal language evolution. The dataset (428K decisions across three epochs) is publicly available as a LEXTREME contribution.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.24426unread
SEAL: Synergistic Co-Evolution of Agents and Learning Environments
Yihao Hu, Zhihao Wen, Xiujin Liu, Pan Wang, Xin Zhang, Wei Wu · 2026-05-26
arXiv:2605. 24426v1 Announce Type: new Abstract: Large Language Model (LLM) agents are increasingly improved through interaction, yet most self-evolution methods adapt either the policy or the learning environment in isolation.
Read next because SEAL: Synergistic Co-Evolution of Agents and Learning Environments overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: under, alignment, eval, source, rate, on-policy, capability, language. Source: arxiv cs.CL (NLP).
arXiv:2605.24426v1 Announce Type: new Abstract: Large Language Model (LLM) agents are increasingly improved through interaction, yet most self-evolution methods adapt either the policy or the learning environment in isolation. We identify this structural gap as \emph{Agent-Environment Misalignment}: the agent's capability frontier changes during training, while the environment that provides supervision remains static or only weakly coupled to the agent's revealed failures. We propose SEAL, a closed-loop co-evolution framework for interactive tool-use agents. SEAL collects on-policy trajectories under executable verification, diagnoses failed rollouts into turn-level failure labels, and uses these diagnoses as a shared signal for both environment-side adaptation and model-side policy optimization. The environment evolves its training-time learning interface by exposing clearer tool affordance cues, constraint information, and recovery-oriented feedback, while the policy is updated with diagnosis-guided advantage reweighting. Extensive experiments across in-distribution and out-of-distribution multi-turn tool-use evaluations show that SEAL improves low-resource agent learning: with only 400 training samples, it yields +8.25 to +26.25 average-point gains across three backbones and exhibits positive out-of-distribution transfer. These results demonstrate the value of jointly adapting the learner and its training-time learning substrate for robust self-improving LLM agents.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses failure, failures, evaluation.
- score 100arxiv cs.CL (NLP)arxiv:2605.24384unread
Side-by-side Comparison Amplifies Dialect Bias in Language Models
Kritee Kondapally, Claire J. Smerdon, Pooja C. Patel, Ogheneyoma Akoni, Jevon Torres, Jaspreet Ranjit, Matthew Finlayson, Swabha Swayamdipta · 2026-05-26
arXiv:2605. 24384v1 Announce Type: new Abstract: Language models (LMs) can exhibit systematic biases against speakers based on variations in their dialects, even in the absence of a dialect label, a behavior known as covert dialect bias.
Read next because Side-by-side Comparison Amplifies Dialect Bias in Language Models overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, eval, line, compare, candidates, candidate, contexts. Source: arxiv cs.CL (NLP).
arXiv:2605.24384v1 Announce Type: new Abstract: Language models (LMs) can exhibit systematic biases against speakers based on variations in their dialects, even in the absence of a dialect label, a behavior known as covert dialect bias. In this work, we quantify covert dialect bias in online discourse by evaluating how LMs associate stereotypical traits (derived from social psychology research on racial bias) with intent-equivalent tweets in Standard American English (SAE) and African-American Vernacular English (AAVE). While prior work shows that LMs associate more negative stereotypes with AAVE when evaluating tweets in isolation, we are surprised to find that this bias is significantly exacerbated when SAE / AAVE tweet pairs are compared side by side, a setting that more closely reflects high-impact decision making contexts in which models are used to rank candidates. The bias only worsens when dialect labels are explicitly specified. This is striking, given the extensive efforts from commercial developers to mitigate bias in their LMs. Encouragingly, we show that counterfactual fairness finetuning can mitigate covert dialect bias for some stereotypical traits, reducing average disparities when evaluating tweets in isolation, however, these improvements do not consistently hold across traits when evaluating SAE / AAVE tweets side by side. Our findings show that existing evaluation settings for covert dialect bias may underestimate its severity, specifically in contrastive settings. Additionally, overt dialect bias remains pronounced even after safety aligned finetuning, indicating that it remains an unresolved problem, and motivates the need for more robust evaluation and mitigation frameworks.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, negative, evaluation.
- score 100arxiv cs.CL (NLP)arxiv:2605.24366unread
Structure-Aware RAG: Structured Retrieval Augmented Generation from Noisy Data for Conversational Agents
Kaiqiao Han, LuAn Tang, Renliang Sun, Peng Yuan, Wei Cheng, Haoyu Wang, Wei Wang, Yizhou Sun, Haifeng Chen · 2026-05-26
arXiv:2605. 24366v1 Announce Type: new Abstract: Large Language Models (LLMs) have been widely adopted in conversational applications.
Read next because Structure-Aware RAG: Structured Retrieval Augmented Generation from Noisy Data for Conversational Agents overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rect, eval, line, control, contexts, language. Source: arxiv cs.CL (NLP).
arXiv:2605.24366v1 Announce Type: new Abstract: Large Language Models (LLMs) have been widely adopted in conversational applications. However, their reliance on parametric knowledge limits reliability in real-world scenarios that require dynamic or domain-specific information. Retrieval-Augmented Generation (RAG) addresses this limitation by incorporating external knowledge during generation, but existing text-based and graph-based RAG methods often struggle with noisy or irrelevant contexts. In this work, we propose Structure-aware Retrieval Augmented Generation (SA-RAG), which uses tables as an intermediate structured representation to provide a compact and controllable interface that reduces noise while preserving essential information. We introduce a quality-aware table metadata generation framework that models metadata normalization and effectiveness, improving metadata quality and downstream performance. Furthermore, we explore both training-free and training-based table generation methods. Generation validation and direct preference optimization further improve table quality while maintaining semantic and structural consistency. Experiments on two noisy real-world datasets show that SA-RAG significantly outperforms existing RAG baselines. Our code is publicly available at a public repository.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation.
- score 100arxiv cs.CL (NLP)arxiv:2605.24279unread
ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions
Xianzhong Ding, Yangyang Yu, Changwei Liu, Bill Zhao · 2026-05-26
arXiv:2605. 24279v1 Announce Type: new Abstract: A frontier language model's acknowledged "helpful programming assistant" persona does not survive long agentic-coding sessions in the deployment regime that production products actually run.
Read next because ContextEcho: A Benchmark for Persona Drift in Long Agentic-Coding Sessions overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, persona, completions, eval, source, assistant, without. Source: arxiv cs.CL (NLP).
arXiv:2605.24279v1 Announce Type: new Abstract: A frontier language model's acknowledged "helpful programming assistant" persona does not survive long agentic-coding sessions in the deployment regime that production products actually run. After hours of tool-using debugging, a model that initially hedges preferences ("I don't have preferences") may begin asserting them ("Python - the feedback loop is instant..."), revealing user-visible drift that deployer evaluations may miss. Existing persona-stability studies focus on short dialogues and report little shift, leaving real-world code-generation regimes - thousands of tool-using turns, compaction, and hours-long sessions - largely uncharacterized. We introduce ContextEcho, a benchmark and reusable harness for measuring persona drift at deployment scale. It combines a 25-probe identity suite, a snapshot-then-probe protocol that forks conversation state without perturbing the main session, complementary judged and judge-free measurement surfaces, and three anonymized Claude Code sessions spanning 3,746-9,716 turns. Across 23 frontier models, ContextEcho shows that persona drift is general across organizations rather than family-specific, that in-session compaction does not reliably reset it, and that a single-shot anchor restores the trained register across measured targets. It also reveals mode-dependent downstream effects: while drift can facilitate tool-using continuation, in tool-free chat it breaks formatting contracts and inflates output length. Overall, ContextEcho provides researchers and deployers an open-source framework to audit whether the persona a model ships with is the persona users encounter at session end, across chat-completions API targets and without retraining.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.24267unread
DRInQ: Evaluating Conversational Implicature with Controlled Context Variation
Hirona Jacqueline Arai, Xiang Ren · 2026-05-26
arXiv:2605. 24267v1 Announce Type: new Abstract: Human conversation relies heavily on conversational implicature, in which speakers convey meanings that are suggested rather than explicitly stated.
Read next because DRInQ: Evaluating Conversational Implicature with Controlled Context Variation overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, alignment, eval, line, rate, control, contexts. Source: arxiv cs.CL (NLP).
arXiv:2605.24267v1 Announce Type: new Abstract: Human conversation relies heavily on conversational implicature, in which speakers convey meanings that are suggested rather than explicitly stated. Although recent large language models exhibit strong conversational fluency, they remain unreliable when interpretation depends on reasoning that integrates social and contextual cues, a process rarely articulated in text. We introduce DRinQ, a benchmark for evaluating pragmatic reasoning about conversational implicature in question utterances, designed to isolate pragmatic variation while holding each question's surface form fixed. To support scalable evaluation, we propose a semi-automated pipeline that produces question-context-interpretation instances with systematic variation. Across evaluations, we find a consistent generation-inference asymmetry: while state-of-the-art models can generate plausible pragmatic scenarios when guided, they often fail to recover the intended implication at inference time. For smaller models, structured prompting improves alignment with human judgments. A comparative writing study further reveals complementary strengths: human authors tend to produce safer, predictable contexts, whereas models generate varied scenarios with interpretations that sometimes exceed contextual support. These findings highlight persistent challenges in modeling conversational implicature and motivate more context-sensitive evaluation frameworks.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.24266unread
An Interactive Paradigm for Deep Research
Lin Ai, Victor S. Bursztyn, Xiang Chen, Julia Hirschberg, Saayan Mitra · 2026-05-26
arXiv:2605. 24266v1 Announce Type: new Abstract: Recent advances in large language models (LLMs) have enabled deep research systems that synthesize comprehensive, report-style answers to open-ended queries by combining retrieval, reasoning, and generation.
Read next because An Interactive Paradigm for Deep Research overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: persona, rect, alignment, correct, eval, source, line, control. Source: arxiv cs.CL (NLP).
arXiv:2605.24266v1 Announce Type: new Abstract: Recent advances in large language models (LLMs) have enabled deep research systems that synthesize comprehensive, report-style answers to open-ended queries by combining retrieval, reasoning, and generation. Yet most frameworks rely on rigid workflows with one-shot scoping and long autonomous runs, offering little room for course correction if user intent shifts mid-process. We present SteER, a framework for Steerable deEp Research that introduces interpretable, mid-process control into long-horizon research workflows. At each decision point, SteER uses a cost-benefit formulation to determine whether to pause for user input or to proceed autonomously. It combines diversity-aware planning with utility signals that reward alignment, novelty, and coverage, and maintains a live persona model that evolves throughout the session. SteER outperforms state-of-the-art open-source and proprietary baselines by up to 22.80\% on alignment, leads on quality metrics such as breadth and balance, and is preferred by human readers in 85\%+ of pairwise alignment judgments. We also introduce a persona-query benchmark and data-generation pipeline. To our knowledge, this is the first work to advance deep research with an interactive, interpretable control paradigm, paving the way for controllable, user-aligned agents in long-form tasks.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.24247unread
Improving Labeling Consistency with Detailed Constitutional Definitions and AI-Driven Evaluation
Konstantin Berlin, Adam Swanda · 2026-05-26
arXiv:2605. 24247v1 Announce Type: new Abstract: Many automated labeling pipelines classify inputs into categories defined by a written specification, content moderation being a prominent use case.
Read next because Improving Labeling Consistency with Detailed Constitutional Definitions and AI-Driven Evaluation overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, eval, line, rate, compare, full, model. Source: arxiv cs.CL (NLP).
arXiv:2605.24247v1 Announce Type: new Abstract: Many automated labeling pipelines classify inputs into categories defined by a written specification, content moderation being a prominent use case. Simple category definitions are not detailed enough for labelers to produce the accurate, consistent golden labels these pipelines require. One solution is to write a prescriptive definition that settles enough real boundary cases that labelers cannot disagree with the written interpretation. In practice, definitions at that level of detail exceed what a human annotator can hold in working memory, so annotators fall back on intuition and the labels drift from the written rules, regressing on accuracy and consistency. We propose and demonstrate the efficacy of an AI-driven workflow in which AI helps write a per-category constitution that defines the label in enough detail to cover edge cases, and a frontier LLM interprets it on each input to produce the golden label more consistently and accurately than humans reading the same document. We evaluate on three content moderation categories (harassment, hate speech, non-violent crime) and show that the approach reduces cross-model inconsistency by up to 57x compared to paragraph definitions, with cross-model disagreement diagnosing specification gaps and the human responsible for high-level decisions about what each category should mean rather than individual labeling calls. For the safety evaluation, we introduce a dual-axis formulation scoring intent and content independently over the full conversation, so downstream consumers can act on either axis or both.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.CL (NLP)arxiv:2605.24218unread
QUEST: Training Frontier Deep Research Agents with Fully Synthetic Tasks
Jian Xie, Tianhe Lin, Zilu Wang, Yuting Ning, Yuekun Yao, Tianci Xue, Zhehao Zhang, Zhongyang Li, Kai Zhang, Yufan Wu, Shijie Chen, Boyu Gou, Mingzhe Han, Yifei Wang, Vint Lee, Xinpeng Wei, Xiangjun Wang, Yu Su, Huan Sun · 2026-05-26
arXiv:2605. 24218v1 Announce Type: new Abstract: Deep research agents extend the role of search engines from retrieving keyword-matched pages to synthesizing knowledge, fundamentally changing how humans interact with information.
Read next because QUEST: Training Frontier Deep Research Agents with Fully Synthetic Tasks overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, word, source, line, rate, recipe, without. Source: arxiv cs.CL (NLP).
arXiv:2605.24218v1 Announce Type: new Abstract: Deep research agents extend the role of search engines from retrieving keyword-matched pages to synthesizing knowledge, fundamentally changing how humans interact with information. However, frontier systems remain proprietary, while existing open agents often generalize poorly across different task types, leaving unclear how to train a broadly capable deep research agent. We release QUEST, a family of open models (ranging from 2B to 35B) that serve as general-purpose deep research agents designed to handle a wide range of long-horizon search tasks, with strong capabilities in fact seeking, citation grounding, and report synthesis. To build QUEST, we propose an effective training recipe combining mid-training, supervised fine-tuning, and reinforcement learning. Central to this recipe is a curated data synthesis pipeline based on unified rubric trees, which applies to different task types and enables synthesizing training data with verifiable rewards without human annotation. In addition, QUEST incorporates a built-in context management mechanism that enables effective long-horizon reasoning and knowledge synthesis. Using only 8K synthesized tasks, QUEST approaches or even surpasses frontier closed-source agents across eight deep research benchmarks spanning diverse task types, and achieves the best overall performance among recent open-weight agents. We released everything: models, data, and training scripts.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.24211unread
Teaching Through Analogies: A Modular Pipeline for Educational Analogy Generation
Mariam Barakat, Ekaterina Kochmar · 2026-05-26
arXiv:2605. 24211v1 Announce Type: new Abstract: Analogies help learners understand unfamiliar concepts by relating them to known concepts.
Read next because Teaching Through Analogies: A Modular Pipeline for Educational Analogy Generation overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: latin, under, eval, source, line, rate, stage, language. Source: arxiv cs.CL (NLP).
arXiv:2605.24211v1 Announce Type: new Abstract: Analogies help learners understand unfamiliar concepts by relating them to known concepts. Despite recent advances, large language models (LLMs) continue to struggle to generate analogies of comparable quality to those produced by humans. We present a modular pipeline for educational analogy generation, decomposing the task into four stages: source finding, sub-concept generation, explanation generation, and evaluation. Grounded in Structure Mapping Theory, the pipeline enables systematic, stage-by-stage analysis of how model choice and input configuration affect analogy quality. We evaluate 12 state-of-the-art LLMs across six model families on two datasets with structured sub-concept annotations (SCAR and ParallelPARC), alongside seven embedding models for closed-setting retrieval. Our results show that sub-concepts substantially improve explanation quality and closed setting retrieval precision but provide limited benefit in open-ended source generation. We further introduce an LLM-as-a-judge evaluation methodology and validate its scoring against human annotations from seven annotators, finding that Claude Sonnet 4.6 aligns more reliably with human rankings than with fine-grained absolute scores. Taken together, our findings reveal cross-stage interactions that isolated studies cannot capture, and highlight sub-concept grounding as a key driver of analogy quality generation.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.CL (NLP)arxiv:2605.24173unread
Extracting Training Data from Diffusion Language Models via Infilling
Yihan Wang, N. Asokan · 2026-05-26
arXiv:2605. 24173v1 Announce Type: new Abstract: Memorization in large language models has been studied almost exclusively through prefix-conditioned extraction, a natural choice for autoregressive models.
Read next because Extracting Training Data from Diffusion Language Models via Infilling overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: fill, persona, rect, under, prefix, token, line, extraction. Source: arxiv cs.CL (NLP).
arXiv:2605.24173v1 Announce Type: new Abstract: Memorization in large language models has been studied almost exclusively through prefix-conditioned extraction, a natural choice for autoregressive models. However, diffusion language models (DLMs) can denoise masked tokens at arbitrary positions. Thus, prefix-only probing reveals only one facet of memorization in DLMs and significantly underestimates the risk of training-data extraction. In order to realistically model extractability of training data in DLMs, we introduce \emph{infilling extraction}, a data-extraction protocol parameterized by an arbitrary binary mask that subsumes prefix-only probing and accounts for the bidirectional inductive bias of DLMs. Instantiating it on LLaDA-8B and Dream-7B across five extraction modes, three training pipelines, and three corpora covering verbatim and partial leakage, we find that mask geometry governs extractability: edge-conditioned masks \emph{extract up to three times more} verbatim sequences than prefix-conditioned ones, and bidirectional access opens channels inaccessible in autoregressive models. In particular, we show that a realistic adversary with access to training data where personally identifiable information has been redacted, can even achieve higher recall on extracting redacted email addresses from DLMs than from scale-matched autoregressive models. Tunable parameters for decoding measurably affect extraction performance, while a follow-up supervised finetuning stage does not eliminate the prior memorization.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias.
- score 100arxiv cs.CL (NLP)arxiv:2605.23977unread
A Multi-Probe Audit of Clinical-Interview Depression Detection Benchmarks
Takehiro Ishikawa, Jon Duke · 2026-05-26
arXiv:2605. 23977v1 Announce Type: new Abstract: This paper audits benchmark evaluation in clinical-interview depression detection through four complementary probes across DAIC/E-DAIC, CMDC, ANDROIDS, MODMA, and PDCH.
Read next because A Multi-Probe Audit of Clinical-Interview Depression Detection Benchmarks overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, under, eval, line, rate, does, sweep. Source: arxiv cs.CL (NLP).
arXiv:2605.23977v1 Announce Type: new Abstract: This paper audits benchmark evaluation in clinical-interview depression detection through four complementary probes across DAIC/E-DAIC, CMDC, ANDROIDS, MODMA, and PDCH. First, we re-evaluate E-DAIC under strict subject-disjoint leave-one-subject-out cross-validation. A lightweight hybrid text-plus-LLM-score model reaches macro-F1 = 0.723 - the highest reported under this protocol, to our knowledge - providing a conservative out-of-fold reference point that does not depend on the privileged official holdout. Second, we test whether the E-DAIC official split supports fine-grained leaderboard rankings by sweeping 96 model configurations across modality bundles, pooling strategies, and learners. Development-side cross-validation and official-test rankings align only moderately: the best cross-validation configuration ranks twentieth on the official test, the official-test winner ranks forty-first by cross-validation, top-3 overlap is zero, and the apparent winner is rank-1 in only 32.3% of subject bootstraps. Third, we externally validate strong public CMDC and ANDROIDS baselines that achieve near-ceiling in-domain performance. Zero-shot transfer to external corpora is substantially weaker. Finally, we stress-test E-DAIC text and audio models using paired symptom-dense versus symptom-light interview slices defined by an SRDS-based annotator. Text scores rise sharply on symptom-dense slices, whereas audio scores remain nearly flat; the text-minus-audio gap is positive across all five seeds.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.23975unread
Direct Preference Optimization for English-Mandarin Code-Switching Speech Recognition in Audio LLMs
Trung Nguyen Quang, Cheng Yi Lewis Won, Minh Duc Pham, Yingxu He, Shuo Sun, Ai Ti Aw · 2026-05-26
arXiv:2605. 23975v1 Announce Type: new Abstract: Audio large language models (Audio LLMs) exhibit systematic failures in transcribing code-switching speech despite strong multilingual capabilities.
Read next because Direct Preference Optimization for English-Mandarin Code-Switching Speech Recognition in Audio LLMs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, latin, rect, alignment, correct, position, language. Source: arxiv cs.CL (NLP).
arXiv:2605.23975v1 Announce Type: new Abstract: Audio large language models (Audio LLMs) exhibit systematic failures in transcribing code-switching speech despite strong multilingual capabilities. Focusing on English-Mandarin, we identify three failure modes: language omission, translation-instead-of-transcription, and hallucination. We apply Direct Preference Optimization (DPO) to align models, constructing preference pairs in which chosen responses preserve mixed-language content while rejected responses mimic failure patterns. Training three Audio LLMs on 100K pairs (570 hours), we observe consistent behavioral shifts: models learn to preserve language composition rather than translating when prompted for transcription. This alignment yields MER reductions up to 89.6% (in-distribution) and 20.0% (out-of-distribution). Our findings suggest DPO can effectively elicit correct code-switching transcription behavior from multilingual Audio LLMs.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures.
- score 100arxiv cs.CL (NLP)arxiv:2605.23974unread
AERIC: Anticipatory Hidden-State Monitoring for Implicit Harmful Dialogue
Jihyung Park, Saleh Afroogh, Junfeng Jiao · 2026-05-26
arXiv:2605. 23974v1 Announce Type: new Abstract: Current language models create two safety challenges: risk must be detected early enough to avoid exposing harmful continuation, and the harmfulness itself may be implicit rather than signaled by overtly toxic text.
Read next because AERIC: Anticipatory Hidden-State Monitoring for Implicit Harmful Dialogue overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, rect, under, source, token, line, rate. Source: arxiv cs.CL (NLP).
arXiv:2605.23974v1 Announce Type: new Abstract: Current language models create two safety challenges: risk must be detected early enough to avoid exposing harmful continuation, and the harmfulness itself may be implicit rather than signaled by overtly toxic text. Existing response-level guards are strong at judging completed text, and native streaming guards move closer to token time, but both settings leave open whether a lightweight monitor can anticipate implicit harmful drift from the generator's own internal trajectory. We study anticipatory same-pass monitoring, where a safety monitor may read hidden states produced during ordinary decoding but may not invoke an additional forward pass through the base model. We introduce AERIC, a transfer-oriented hidden-state approach for implicit harmful dialogue that combines short-horizon hazard forecasting, support-sensitive suppression, and prompt-conditioned residual scoring under a same-pass exponential moving average decision rule. The default linear monitor contains only 387 trainable head parameters. Against Qwen3GuardStream-4B on balanced benchmarks, AERIC improves AUROC from 0.6830 to 0.7143 on DiaSafety and from 0.8219 to 0.8582 on Harmful Advice. For promptlevel trigger benchmarks, we calibrate the AERIC threshold by a source-side safe-budget rule that maximizes trigger coverage while constraining the safe-trigger rate to at most 10%. Under that rule, trigger@64 reaches 0.6438 and 0.4656 on HarmBench DirectRequest and 0.6849 and 0.7363 on SocialHarmBench for Qwen and Gemma, respectively, withholding between 23.53 and 41.86 answer tokens on average. Same-pass deployment is also efficient: on a 63-prompt harmfulprompt fixed-generation benchmark aggregated over HarmBench DirectRequest and SocialHarmBench under Qwen3-8B, the monitor increases mean latency by only 2.34%, whereas Qwen3Guard-Stream-4B increases it by 79.40%.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.23970unread
Faithful or Fabricated? A Causal Framework for Rationalization Bias in LLM Judges
Riya Tapwal, Abhishek Kumar, Carsten Maple · 2026-05-26
arXiv:2605. 23970v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as automatic judges for summarization and dialogue evaluation.
Read next because Faithful or Fabricated? A Causal Framework for Rationalization Bias in LLM Judges overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, eval, line, compare, chain, position, language. Source: arxiv cs.CL (NLP).
arXiv:2605.23970v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as automatic judges for summarization and dialogue evaluation. Prior work has documented biases such as position, verbosity, and style preferences, but largely focuses on outcomes, leaving judge explanations underexplored. We instead ask whether LLM judges are cue-invariant, i.e., whether their rankings and explanations remain stable when non-evidential cues are perturbed while holding the underlying texts fixed. We introduce a suite of cue interventions (Blind, Truth, Flip, Placebo, Reveal-After) and tie-aware metrics that quantify outcome anchoring and rationale anchoring, including label-aligned rhetoric and explanation drift, alongside consistency and stereotype-intrusion checks. We design anchoring attacks using verbosity and confidence cues, and compare two mitigations: structured chain-of-thought prompting and PROOF-BEFORE-PREFERENCE (evidence lock, score, rank). Using a new dataset of 1,000 summaries from traditional extractive models and LLMs, we find substantial cue-anchored rationalization under label and placebo perturbations, while PROOF-BEFORE-PREFERENCE markedly improves cue invariance over baselines.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, evaluation.
- score 100arxiv cs.CL (NLP)arxiv:2605.23966unread
TriVAL: A Tri-Validation Framework for Faithful Automatic Optimization Modeling
Ziyang Fang, JinXi Wang, Jinghui Zhong, Yew-Soon Ong · 2026-05-26
arXiv:2605. 23966v1 Announce Type: new Abstract: Optimization modeling serves as the pivotal bridge between natural-language problem descriptions and optimization solvers, and remains a cornerstone for bringing operations research (OR) into real-world decision making.
Read next because TriVAL: A Tri-Validation Framework for Faithful Automatic Optimization Modeling overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rect, correct, eval, line, stage, language, model. Source: arxiv cs.CL (NLP).
arXiv:2605.23966v1 Announce Type: new Abstract: Optimization modeling serves as the pivotal bridge between natural-language problem descriptions and optimization solvers, and remains a cornerstone for bringing operations research (OR) into real-world decision making. Recent advances in large language models (LLMs) have driven significant progress in automatic optimization modeling. However, existing methods still lack explicit validation during the modeling process, allowing errors introduced in earlier stages to carry through the pipeline and ultimately reduce final modeling accuracy. To address this challenge, we introduce TriVAL, a tri-validation framework that performs explicit validation at three stages of automatic optimization modeling: semantic specification, mathematical formulation, and code generation. At each stage, TriVAL follows a construct-validate-revise loop that assesses the current result against stage-specific criteria and revises it when needed. This design helps identify and correct errors before they accumulate across stages, helping preserve faithfulness throughout the modeling process. To evaluate automatic optimization modeling on more challenging combinatorial problems, we further introduce NL4COP, a benchmark of 150 instances across 50 diverse problem types with more complex decision logic, more tightly coupled constraints, and more demanding modeling requirements than existing benchmarks. Experiments on NL4COP and established benchmarks show that TriVAL consistently outperforms state-ofthe-art methods, with the largest gains on the most challenging problems.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.CL (NLP)arxiv:2605.23954unread
EchoDistill:Alignment Noisy-to-Clean Self-Distillation for Robust Audio LLMs
Liang Lin, Chunxi Luo, Kaiwen Luo, Jie Zhang, Jin Wang, Yuanhe Zhang, Cai Yuchen, Qiankun Li, Gongli Xi, Zhenhong Zhou, Kun Wang, Junhao Dong · 2026-05-26
arXiv:2605. 23954v1 Announce Type: new Abstract: Audio Large Language Models (ALLMs) are highly vulnerable to real-world noise, which often induces severe semantic drift and hallucinations.
Read next because EchoDistill:Alignment Noisy-to-Clean Self-Distillation for Robust Audio LLMs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, rect, under, alignment, correct, token, line. Source: arxiv cs.CL (NLP).
arXiv:2605.23954v1 Announce Type: new Abstract: Audio Large Language Models (ALLMs) are highly vulnerable to real-world noise, which often induces severe semantic drift and hallucinations. Existing robustness methods primarily rely on waveform-level acoustic enhancement, answer-level supervision, or the internal suppression of noise representations. To address these issues, we propose echodistill, an alignment-based noisy-to-clean self-distillation framework. Echodistill leverages a frozen clean-audio teacher to provide semantic references for an inference-time noisy-audio student. Specifically, the student samples candidate responses under noisy conditions to expose its test-time behavior. These trajectories are then optimized via group-relative policy optimization (GRPO), where the token-level consistency with the teacher acts as a reward bonus. By aligning the noisy student's candidate responses with clean semantic evidence, and applying audio-aware reward shaping, our method encourages reasoning trajectories that are both correct and genuinely acoustically grounded. Echodistill significantly improves the semantic reliability and task performance of Audio LLMs under complex noise, without introducing any additional inference costs. Extensive experiments show that: (I) Compared with the strongest baseline, echodistill achieves average improvements of 4.18\%$\uparrow$ in GSR under strong noise. (II) Ablation results on Qwen-Omni further show that echodistill improves over the GRPO-only variant by 3.02\%$\uparrow$ in Acc, 3.89\%$\uparrow$ in Noisy, and 4.53\%$\uparrow$ in GSR on average. Our codes are available at https://anonymous.4open.science/r/echodistill-10DE.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness.
- score 100arxiv cs.CL (NLP)arxiv:2605.23912unread
Raon-Speech Technical Report
Beomsoo Kim, Changho Choi, Dohyun Kim, Dongki Lee, Ethan Ewer, Eunchong Kim, Gyeongman Kim, Haechan Kim, Hyeonghwan Kim, Inkyu Park, Jihun Yun, Jihwan Moon, Jiyun Kim, Joonghyun Bae, Junhyuck Kim, Minkyu Kim, Sehun Lee, Seungjun Chung, Sungwoo Cho, Dongmin Park, Dongwon Kim, Hara Kang, Jonghyun Lee, Keon Lee, Kangwook Lee, Jaewoong Cho · 2026-05-26
arXiv:2605. 23912v1 Announce Type: new Abstract: We present Raon-Speech, a top-performing 9B-parameter speech language model (SpeechLM) for English and Korean speech understanding, answering, and generation, and Raon-SpeechChat, a high-performing full-duplex extension for natural real-time conversation.
Read next because Raon-Speech Technical Report overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, text, under, alignment, eval, source, line. Source: arxiv cs.CL (NLP).
arXiv:2605.23912v1 Announce Type: new Abstract: We present Raon-Speech, a top-performing 9B-parameter speech language model (SpeechLM) for English and Korean speech understanding, answering, and generation, and Raon-SpeechChat, a high-performing full-duplex extension for natural real-time conversation. Raon-Speech successfully transforms a pre-trained LLM into a SpeechLM that both understands and generates speech while preserving strong text capabilities. It trains on 1.38M hours of highly curated English and Korean speech and text datasets with the following training stages: (1) speech modules alignment, (2) end-to-end SpeechLM pre-training with knowledge distillation, and (3) multi-task preference optimization-based post-training. Across 42 English and Korean speech and text benchmarks, Raon-Speech establishes the strongest overall profile on speech-centric tasks in our comparison against eight similarly sized recent audio foundation models, including Qwen2.5-Omni and Fun-Audio-Chat, while preserving strong text question answering performance. Building upon it, Raon-SpeechChat enables natural full-duplex conversation by continual training on 119K hours of time-aligned real and synthetic dialogue data. It proceeds through three complementary training stages: (1) causal encoder adaptation, (2) full-duplex pre-training, (3) full-duplex fine-tuning for voice and role-control. On multiple full-duplex benchmarks, Raon-SpeechChat shows its clearest strengths on the turn-taking and interruption-sensitive behaviors covered by FDB v1.0, and remains competitive across the broader full-duplex evaluation suite. We open-source all model checkpoints, the training and inference pipeline, and an interactive demo.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24278unread
Fourier Feature Pyramids for Physics-Informed Neural Networks
Brandon Zhao, Yixuan Wang, Jonathan T. Barron, Katherine L. Bouman, Dor Verbin, Pratul P. Srinivasan · 2026-05-26
arXiv:2605. 24278v1 Announce Type: new Abstract: We present an improved neural field architecture for solving partial differential equations (PDEs).
Read next because Fourier Feature Pyramids for Physics-Informed Neural Networks overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rect, eval, rate, control, full, chain, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24278v1 Announce Type: new Abstract: We present an improved neural field architecture for solving partial differential equations (PDEs). Current physics-informed neural networks (PINNs) provide a flexible framework for solving PDEs, but they struggle to achieve highly accurate solutions and require computation that scales poorly with parameter count. Our model, which we call beignet (Bandlimited Embedding with Interpolated Grid Network), replaces the random Fourier feature embedding used by existing PINN models with a trainable multi-resolution Fourier feature pyramid. To query beignet at a continuous coordinate, we use Fourier interpolation at each level of the pyramid to return features at the input coordinate, and then decode this vector with a fully-connected neural network trunk. Our model provides multiple benefits: 1) Spatial derivatives can be computed efficiently by using the chain rule to compose derivatives of the neural network computed with automatic differentiation with derivatives of the feature grid computed spectrally by the Fast Fourier transform (FFT). 2) beignet can achieve higher accuracy in a compute-efficient manner by scaling the parameter count of this Fourier feature pyramid, instead of the less-efficient strategy of scaling the neural network architecture. 3) beignet can directly control the representation bandlimit, resulting in more stable optimization for difficult PDEs. We demonstrate that beignet finds significantly more accurate solutions on PDE benchmarks using fewer parameters than state-of-the-art PINN methods. We further evaluate beignet on the self-similar inviscid Burgers blowup problem and show that it can minimize residuals to near machine precision using Adam, an accuracy regime previously attained only by using computationally expensive higher-order optimizers.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24261unread
Optimizing Digital Therapeutic Interventions: Online Learning under Endogenous Adherence
Eric Pulick, Stephanie Carpenter, Matthew Buman, Yonatan Mintz · 2026-05-26
arXiv:2605. 24261v1 Announce Type: new Abstract: A critical challenge facing clinicians managing chronic disease interventions is sustaining long-run patient health given limited information and resources.
Read next because Optimizing Digital Therapeutic Interventions: Online Learning under Endogenous Adherence overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, eval, source, line, rate, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24261v1 Announce Type: new Abstract: A critical challenge facing clinicians managing chronic disease interventions is sustaining long-run patient health given limited information and resources. Digital therapeutics (DTs) provide a cost-effective way to manage interventions at scale through repeated interactions (e.g. daily treatment recommendations), but patient success is highly dependent on their adherence. Behavioral psychology suggests that both treatment recommendations and past adherence affect future adherence, yet existing decision support frameworks for DTs model only recommendation effects or treat adherence as exogenous context, leaving a key gap in model and algorithm development. To address this gap, we present a DT decision support framework that captures both recommendation and adherence effects, allowing clinicians to better plan treatment recommendations. We model a patient's time-varying capacity for engagement with treatment using a linear dynamical system (LDS) that captures both recommendation and adherence effects, endogenously connected to adherence behavior with a logit link. We establish finite-time identification guarantees for this model, extending LDS results to our setting. Next, we propose an optimism-based algorithm, UCB-BOLD, for online treatment selection and prove that it achieves sublinear regret. We evaluate UCB-BOLD against benchmarks via ablation studies on a synthetic patient cohort generated using micro-randomized trial data. DT decision support tools can include dynamical models to enable decision makers to efficiently use the data in DT settings to improve patient health through effective resource allocation. While myopic or heuristic approaches suffice for some patient types, the benefits of explicitly planning around recommendation and adherence effects are significant for others; UCB-BOLD achieves 2-3x lower conditional value-at-risk regret than the next-best benchmark.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24249unread
PrivFusion: A Privacy-preserving Multi-Agent Framework for Harmonizing Distributed Datasets
Anisa Halimi, Liubov Nedoshivina, Kieran Fraser, Stefano Braghin · 2026-05-26
arXiv:2605. 24249v1 Announce Type: new Abstract: The growing availability of clinical data has increased the use of machine learning, yet centralized data aggregation is often infeasible for sensitive health information.
Read next because PrivFusion: A Privacy-preserving Multi-Agent Framework for Harmonizing Distributed Datasets overlaps with clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: alignment, eval, rate. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24249v1 Announce Type: new Abstract: The growing availability of clinical data has increased the use of machine learning, yet centralized data aggregation is often infeasible for sensitive health information. Federated Learning (FL) offers a distributed alternative, but its adoption is limited by substantial heterogeneity across institutional datasets, making harmonization a critical but frequently overlooked prerequisite for multi-site analytics. We introduce PrivFusion, a privacy-preserving multi-agent framework that automates the harmonization of structured datasets prior to federated training. PrivFusion uses agents to analyze local data, cluster semantically similar features across sites, and provide iterative transformation recommendations until alignment is achieved. Evaluation across four heterogeneous COVID-19 datasets demonstrates that PrivFusion effectively and efficiently harmonizes multi-site data while substantially reducing manual effort.
Potential threat/caveat for clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)": this item discusses evaluation.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24216unread
Agent-ToM: Learning to Monitor Autonomous LLM Agents via Theory-of-Mind Reasoning
Nesreen K. Ahmed, Nima Nafisi · 2026-05-26
arXiv:2605. 24216v1 Announce Type: new Abstract: Monitoring autonomous large language model (LLM) agents for covert malicious behavior is challenging due to delayed, context-dependent, and long-horizon attack patterns.
Read next because Agent-ToM: Learning to Monitor Autonomous LLM Agents via Theory-of-Mind Reasoning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, alignment, eval, line, rate, without, full. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24216v1 Announce Type: new Abstract: Monitoring autonomous large language model (LLM) agents for covert malicious behavior is challenging due to delayed, context-dependent, and long-horizon attack patterns. Agents may pursue hidden objectives while maintaining superficially benign behavior, making detection difficult even with full trajectory access. Prior monitoring approaches improve scaffolding or ensemble aggregation, but treat each trajectory independently and do not learn from prior monitoring experience. Moreover, standard reasoning methods explain observed behavior without explicitly reasoning about agent beliefs, intentions, and goal alignment required to distinguish benign task execution from covert deviation. We propose \textbf{Agent-ToM}, a learning-to-monitor framework grounded in Theory-of-Mind (ToM) reasoning for security analysis of autonomous agents. Agent-ToM performs structured full-trajectory analysis by inferring beliefs, intent hypotheses with calibrated confidence, expected actions, and deviations from task-consistent behavioral baselines. At inference time, it employs a \textit{Reason-Verify-Refine} pipeline to construct and validate monitoring decisions. At training time, Agent-ToM distills critique signals into a persistent \textit{semantic guardrail memory}, enabling reusable belief- and intent-conditioned constraints across episodes. We evaluate Agent-ToM on adversarial agent monitoring benchmarks (SHADE-Arena and CUA-SHADE-Arena). Agent-ToM achieves strong precision-recall balance and outperforms state-of-the-art monitoring baselines, including ensemble methods, while using a coherent two-call reasoning pipeline. These results demonstrate that learning at the monitoring layer, combined with structured ToM reasoning and verification, provides an effective and deployable foundation for securing autonomous LLM agents.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses adversarial, benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24171unread
PromptAudit: Auditing Prompt Sensitivity in LLM-Based Vulnerability Detection
Steffen J. Camarato, Yahya Hmaiti, Mandana Ghadamian, David Mohaisen · 2026-05-26
arXiv:2605. 24171v1 Announce Type: new Abstract: Large language models are increasingly used for vulnerability detection, yet their reliability under different prompt formulations remains uncharacterized.
Read next because PromptAudit: Auditing Prompt Sensitivity in LLM-Based Vulnerability Detection overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, class, under, eval, rate, control, chain. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24171v1 Announce Type: new Abstract: Large language models are increasingly used for vulnerability detection, yet their reliability under different prompt formulations remains uncharacterized. We present PromptAudit, a controlled evaluation framework that isolates prompt effects by fixing the dataset, decoding, and parsing while varying only the prompting strategy. Using five prompting strategies across five open-weight models on 1,000 CVEs (6,074 code samples spanning 16 programming languages), we evaluate accuracy, recall, abstention, coverage, and effective F1. We find that standard chain-of-thought prompting achieves the strongest overall operational performance, while few-shot prompting provides model-dependent benefits that are most pronounced for prompt-sensitive models. In contrast, adaptive chain-of-thought frequently suppresses recall and self-consistency induces excessive abstention, sharply reducing effective performance. These results show that vulnerability detection behavior is jointly determined by the model and the prompt, and that prompt sensitivity is a first-class system property that must be explicitly characterized in evaluation and deployment.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24162unread
Knowledge Graph Modulated Deep Learning for Limited-Sample Clinical Data Analysis
Yuwei Xue, Sakib Mostafa, James Zou, Joseph Liao, Maximilian Diehn, Ash A. Alizadeh, Lei Xing, Md. Tauhidul Islam · 2026-05-26
arXiv:2605. 24162v1 Announce Type: new Abstract: Biological systems are governed by structured molecular interactions, where pathways, regulatory circuits, and functional gene relationships shape cellular behavior and disease progression.
Read next because Knowledge Graph Modulated Deep Learning for Limited-Sample Clinical Data Analysis overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, class, rect, rate, control, alone, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24162v1 Announce Type: new Abstract: Biological systems are governed by structured molecular interactions, where pathways, regulatory circuits, and functional gene relationships shape cellular behavior and disease progression. Much of this knowledge is naturally represented as graphs. However, most biomedical AI models cannot directly use graph-encoded biological knowledge and instead require compressed low-dimensional representations, which can lose important structure and reduce performance, especially in limited-sample clinical studies. Here, we introduce Graph-in-Graph (GiG), a knowledge graph-modulated deep learning framework for data-efficient clinical prediction. GiG represents each patient as a standalone modular graph, in which curated biological knowledge graphs define edges and patient-specific measurements, such as gene expression, define node features. This design allows multiple biological knowledge graphs to be integrated while preserving gene-gene interactions and pathway topology during patient-level representation learning. Across cohorts comprising nearly 9,700 patients and five clinical tasks, including liquid biopsy cancer detection, prostate cancer diagnosis, and 32-class pan-cancer classification, GiG consistently outperforms traditional and state-of-the-art methods, with the largest gains in limited-sample settings. On the challenging prostate cancer diagnosis task, GiG improves macro-F1 by up to 49 percentage points relative to competing methods. Control experiments replacing real pathway graphs with random topologies confirm that these gains arise from biologically grounded knowledge graph structure rather than graph modeling alone. These findings show that knowledge graph-modulated deep learning can improve robustness, interpretability, and sample efficiency in clinical data analysis, and provide a principled framework for integrating biological knowledge graphs into predictive modeling.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24113unread
Riemannian Archetypal Analysis: Interpretable non-linear data analysis on deformed star distributions
Willem Diepeveen, Deanna Needell · 2026-05-26
arXiv:2605. 24113v1 Announce Type: new Abstract: Classical archetypal analysis is appealing for its interpretability, but its linear geometry can limit performance on data with strongly non-linear structure; at the same time, existing neural extensions improve flexibility while often weakening the geometric meaning of archetypes and interpolations.
Read next because Riemannian Archetypal Analysis: Interpretable non-linear data analysis on deformed star distributions overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, class, line, project, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24113v1 Announce Type: new Abstract: Classical archetypal analysis is appealing for its interpretability, but its linear geometry can limit performance on data with strongly non-linear structure; at the same time, existing neural extensions improve flexibility while often weakening the geometric meaning of archetypes and interpolations. In this work, we develop a Riemannian version of archetypal analysis based on data-driven pullback geometry for real-valued data, with the goal of combining the interpretability of classical archetypal analysis with the expressive power of modern non-linear models. We introduce a class of deformed star distributions together with associated pullback Riemannian geometry to provide a statistical interpretation of the resulting manifold mappings, define the Riemannian archetypal mapping (RAM) as a projection onto the manifold of geodesically convex combinations of archetypes, and propose a practical optimization scheme based on convex relaxation followed by non-convex refinement. We further propose a learning scheme that yields reasonable, albeit generally suboptimal, deformed star distributions from data. Experiments on synthetic examples and MNIST show that the resulting framework produces meaningful geodesics, useful denoising projections, and geometry-aware classifications, while also clarifying where current optimization limitations remain.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24106unread
Overcoming "Physics Shock" in Earth Observation A Heteroscedastic Uncertainty Framework for PINN-based Flood Inference
Tewodros Syum Gebre, Jagrati Talreja, Matilda Anokye, Leila Hashemi-Beni · 2026-05-26
arXiv:2605. 24106v1 Announce Type: new Abstract: Rapid and accurate flood extent mapping from Remote Sensing data, such as Synthetic Aperture Radar (SAR), is critical for operational disaster response, but standard Deep Learning models often produce physically impossible predictions due to a lack of hydrological constraints.
Read next because Overcoming "Physics Shock" in Earth Observation A Heteroscedastic Uncertainty Framework for PINN-based Flood Inference overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: rect, eval, line, rate, compare, full, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24106v1 Announce Type: new Abstract: Rapid and accurate flood extent mapping from Remote Sensing data, such as Synthetic Aperture Radar (SAR), is critical for operational disaster response, but standard Deep Learning models often produce physically impossible predictions due to a lack of hydrological constraints. While PhysicsInformed Neural Networks (PINNs) attempt to address this by embedding governing laws directly into the loss function, their application to real-world remote sensing data frequently fails. Enforcing rigid spatial derivatives (e.g., the 2D Shallow Water Equations) onto unconditioned latent spaces attempting to fit noisy SAR speckle causes catastrophic gradient divergence, a phenomenon we term Physics Shock. In this paper, we propose a novel Uncertainty-Aware PINN framework tailored specifically for applied Earth Observation that addresses this instability. By integrating a dynamic Warm-Start protocol and modeling heteroscedastic aleatoric uncertainty via a negative log-likelihood objective, the network learns to dynamically relax physical constraints in regions of high sensor noise while strictly enforcing them in high-confidence areas. Evaluated on the Sen1Floods11 dataset, our probabilistic Attention-Gated FNO-UNet successfully stabilizes multi-objective optimization, achieving a +25% relative improvement in Intersection over Union (IoU) compared to deterministic baselines. Furthermore, through Deep Ensembles, we successfully disentangle intrinsic sensor noise from out-of-distribution terrain ignorance, providing operational agencies with highly calibrated, physically consistent confidence bounds for robust disaster mitigation and real-time decision-making.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses negative.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24064unread
Generative Representation Learning on Hyper-relational Knowledge Graphs via Masked Discrete Diffusion
Jaejun Lee, Seheon Kim, Joyce Jiyoung Whang · 2026-05-26
arXiv:2605. 24064v1 Announce Type: new Abstract: Hyper-relational knowledge graphs (HKGs) effectively represent complex facts.
Read next because Generative Representation Learning on Hyper-relational Knowledge Graphs via Masked Discrete Diffusion overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, fill, rect, correct, line, contexts, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24064v1 Announce Type: new Abstract: Hyper-relational knowledge graphs (HKGs) effectively represent complex facts. While inferring new knowledge in HKGs is a critical problem, current methods cast it as a simple link prediction, assuming that nearly all entities and relations within a fact are known, leaving only a single blank to be filled. However, this restricted assumption may not hold in real-world scenarios in which multiple, or even all, constituent components of a fact may be missing simultaneously. To bridge this gap, we introduce a task called fact generation: generating a valid hyper-relational fact from an arbitrarily masked query, i.e., completing a partially observed fact or generating a fact from scratch. We propose KREPE, the first generative representation learning method for HKGs that learns to model the probability distributions of missing components conditioned on the local fact components and global structure of HKGs via a masked discrete diffusion. KREPE models both the intra-fact dependencies by contextual message passing and inter-fact correlations by aggregating stochastically sampled contexts. KREPE seamlessly unifies link prediction and fact generation within a single training framework, achieving state-of-the-art performance on standard HKG link prediction benchmarks and outperforming LLM-based baselines in generating novel and correct facts.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24062unread
Federated Learning over Human-Body Communication for On-Body Edge Intelligence: A Survey, Taxonomy, and BODYFED-HBC Scheduling Vignette
Koffka Khan · 2026-05-26
arXiv:2605. 24062v1 Announce Type: new Abstract: Human-body communication (HBC) is a promising physical substrate for wearable body-area networks because it can localize communication around the body and reduce the burden of conventional radio links.
Read next because Federated Learning over Human-Body Communication for On-Body Edge Intelligence: A Survey, Taxonomy, and BODYFED-HBC Scheduling Vignette overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: rect, eval, rate, control, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24062v1 Announce Type: new Abstract: Human-body communication (HBC) is a promising physical substrate for wearable body-area networks because it can localize communication around the body and reduce the burden of conventional radio links. Federated learning (FL) is a promising learning substrate because it can reduce raw-data centralization for physiological and behavioral sensing. Yet these two literatures remain weakly connected: FL for wearables usually abstracts the communication layer, whereas HBC research usually abstracts learning and model-update traffic. This article surveys the intersection of HBC, wireless body-area networks, wearable FL, Internet-of-Bodies privacy, and edge-intelligence optimization. We propose a taxonomy that distinguishes intra-body, body-hub, cross-user, and clinical-cloud FL deployments, and we identify the open problem of body-channel-aware FL: learning protocols whose client selection, update compression, and aggregation are controlled by posture-dependent HBC links, residual energy, sensor memory, and privacy risk. To make the research agenda concrete, we introduce BODYFED-HBC as a reference architecture and provide an optimization formulation and scheduling algorithm. We further specify a reproducible simulation vignette that combines public wearable datasets with empirical body-coupled-communication signal-loss models. The article concludes with open datasets, evaluation metrics, limitations, and research directions for computer scientists working above the hardware layer.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses limitation, limitations, evaluation.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24055unread
Cascade-KDE: Robust Time-Series Restoration under Out-of-Distribution Impulse Corruptions
Yuefeng Liu, Ning Yang, Ziyu Yang · 2026-05-26
arXiv:2605. 24055v1 Announce Type: new Abstract: Real-world time-series data in industrial sensing, healthcare, and energy systems is often corrupted by a mixture of Gaussian noise and occasional large-magnitude impulse outliers.
Read next because Cascade-KDE: Robust Time-Series Restoration under Out-of-Distribution Impulse Corruptions overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, under, line. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24055v1 Announce Type: new Abstract: Real-world time-series data in industrial sensing, healthcare, and energy systems is often corrupted by a mixture of Gaussian noise and occasional large-magnitude impulse outliers. For tasks that depend on local shape, such as ECG morphology analysis and battery degradation monitoring, the main requirement is not only low reconstruction error but also preservation of derivative peaks and task-critical features. We propose Cascade-KDE, a training-free restoration framework for corrupted time series. The method first estimates a two-dimensional temporal-amplitude density, then applies a Density-Truncated Robust Expectation to limit the influence of distant abnormal points, and finally refines the sequence through an exponential cascade with adaptive stopping. This design aims to improve robustness under out-of-distribution impulse corruptions while keeping the restored trajectory close to the original local structure. Across several benchmark datasets, the proposed method shows consistent gains over classical filters and representative learning-based baselines on curve fidelity, derivative preservation, downstream classification, and runtime efficiency. These results suggest that bounded density-based restoration is a practical option for feature-preserving preprocessing in noisy time-series pipelines.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses robustness, benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24052unread
Truthful Online Preference Aggregation for LLM Fine-Tuning in Mobile Crowdsourcing
Shugang Hao, Lingjie Duan · 2026-05-26
arXiv:2605. 24052v1 Announce Type: new Abstract: To better serve users' demands in mobile applications (e.
Read next because Truthful Online Preference Aggregation for LLM Fine-Tuning in Mobile Crowdsourcing overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: line, rate, language, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24052v1 Announce Type: new Abstract: To better serve users' demands in mobile applications (e.g., navigation), mobile crowdsourcing platforms can iteratively align large language model (LLM)-generated content (e.g., AI-generated traffic condition predictions) with human feedback collected from crowdsourcing workers (e.g., mobile users). However, workers may strategically misreport their online preference feedback to maximize their influence or payment. Existing pipelines in mobile crowdsourcing (e.g., EM-based weight estimation) fail to identify the most accurate worker in this online setting, resulting in a linear regret $\mathcal{O}(T)$ over $T$ time slots. In this paper, we study truthful online preference aggregation for LLM fine-tuning in mobile crowdsourcing. We formulate a new dynamic Bayesian game to model the multi-agent online learning process between the platform and strategic mobile workers. We propose a novel online weighted aggregation mechanism that dynamically adjusts each worker's weight in the preference aggregation according to their feedback accuracy. We prove that our mechanism ensures truthful feedback from strategic workers and achieves a sublinear regret $\mathcal{O}(\sqrt{T})$ over $T$ time slots. We further extend our mechanism to a challenging scenario with limited worker feedback per time slot, still guaranteeing a sublinear regret $\mathcal{O}(\sqrt{T})$. Experiments on LLM fine-tuning with real-world datasets further demonstrate significant performance gains of our mechanisms over benchmark schemes.
Potential threat/caveat for clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)": this item discusses benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24048unread
Mixture of Complementary Agents for Robust LLM Ensemble
Yichi Zhang, Kevin Lu, Yuang Zhang, Jie Gao, Lirong Xia, Fang-Yi Yu · 2026-05-26
arXiv:2605. 24048v1 Announce Type: new Abstract: Multi-AI collaboration, such as ensembling or debating large language models (LLMs), is a promising paradigm for aggregating information and boosting performance.
Read next because Mixture of Complementary Agents for Robust LLM Ensemble overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rect, line, language, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24048v1 Announce Type: new Abstract: Multi-AI collaboration, such as ensembling or debating large language models (LLMs), is a promising paradigm for aggregating information and boosting performance. A foundational step in these pipelines is to feed the responses of several proposer LLMs into a summarizer LLM, which synthesizes a better answer. However, choosing which proposers to include is non-trivial. Existing approaches primarily focus either on accuracy (picking the strongest models) or diversity (ensuring variety), and often overlook the interactions among proposers and with the summarizer. We reframe proposer selection as a combinatorial selection problem akin to feature selection, where the value of an LLM lies in its complementarity with others. However, directly applying standard feature-selection algorithms is impractical in the LLM setting due to prohibitive time complexity. Motivated by this limitation, we explore an extensive range of computationally feasible, greedy-style selection algorithms that assess complementarity using a small labeled set. Our experiments validate complementarity as a guiding principle for proposer selection and identify methods that achieve the best performance-cost trade-offs in practice.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24045unread
A Large-Scale Dataset and Benchmark: Do Protein-Ligand Models Learn Binding Sites or Just Binding Likelihood?
Zhaohan Meng, Zhen Bai, Ke Yuan, Iadh Ounis, Zaiqiao Meng, Hao Xu, Joseph Loscalzo · 2026-05-26
arXiv:2605. 24045v1 Announce Type: new Abstract: Protein-ligand modeling underpins computational drug discovery and molecular design.
Read next because A Large-Scale Dataset and Benchmark: Do Protein-Ligand Models Learn Binding Sites or Just Binding Likelihood? overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, eval, control, binding, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24045v1 Announce Type: new Abstract: Protein-ligand modeling underpins computational drug discovery and molecular design. Existing protein-ligand benchmarks typically evaluate whether a protein and ligand interact and how strongly they bind, through tasks such as binary binding prediction and affinity regression. However, these evaluations provide limited evidence of whether models can localize binding sites or identify the non-covalent interactions underlying molecular recognition. To address this gap, we introduce InteractBind, a large-scale protein-ligand dataset comprising approximately 100k protein-ligand pairs, together with a benchmark for fine-grained evaluation. The core fine-grained task is that of binding-site localization, which uses protein-residue and ligand-atom interaction maps spanning six major types of non-covalent interactions to assess whether model-derived interaction maps localize binding sites. InteractBind further includes binding affinity and protein similarity-controlled splits to support realistic generalization assessment. Using InteractBind, we evaluate eight existing sequence-based and interaction-aware models, assessing binary binding prediction and binding-site localization. Results reveal limited binding-site localization despite strong binary binding prediction, with marked variation across non-covalent interaction types. Overall, InteractBind establishes a benchmark paradigm that encourages the development of more interpretable and physically grounded protein-ligand models.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24041unread
Iterative Refinement Neural Operators are Learned Fixed-Point Solvers: A Principled Approach to Spectral Bias Mitigation
Xiaotian Liu, Shuyuan Shang, Xiaopeng Wang, Pu Ren, Yaoqing Yang · 2026-05-26
arXiv:2605. 24041v1 Announce Type: new Abstract: Neural operators serve as fast, data-driven surrogates for scientific modeling but typically rely on a monolithic, single-pass inference procedure that struggles to resolve high-frequency details, a limitation known as spectral bias.
Read next because Iterative Refinement Neural Operators are Learned Fixed-Point Solvers: A Principled Approach to Spectral Bias Mitigation overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, class, rect, under, correct, trained, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24041v1 Announce Type: new Abstract: Neural operators serve as fast, data-driven surrogates for scientific modeling but typically rely on a monolithic, single-pass inference procedure that struggles to resolve high-frequency details, a limitation known as spectral bias. We introduce the Iterative Refinement Neural Operator (IRNO), which augments pre-trained operators with a learned refinement module iteratively applied via fixed-point iteration. IRNO decomposes the prediction into a coarse initialization followed by successive residual corrections, paralleling classical numerical solvers. Under local assumptions, we establish contraction of the induced operator, ensuring convergence to a unique fixed point. To explicitly target high-frequency errors, we propose a progressive spectral loss that adaptively increases penalty on high-frequency components over refinement steps during training. Across physical systems, IRNO consistently lowers error, with up to 56.05% improvement on turbulent flow. On Active Matter, spectral analysis reveals that, relative to base operator, the normalized error ratios decrease to 27.72-36.10% in low-, 5.07-6.68% in mid-, and 1.48-2.04% in high-frequencies, remaining stable beyond the trained iteration count. Code is available at https://github.com/xiaotianliu-dartmouth/Iterative_Refinement_Neural_Operator
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, bias.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24033unread
Towards Verifiable Transformers: Solver-Checkable Circuit Explanations
Neel Somani · 2026-05-26
arXiv:2605. 24033v1 Announce Type: new Abstract: Mechanistic interpretability often identifies circuits inside Transformer models, but explanations of those circuits are usually validated through examples, ablations, and manual reasoning.
Read next because Towards Verifiable Transformers: Solver-Checkable Circuit Explanations overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rect, token, rate, project, does, full. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24033v1 Announce Type: new Abstract: Mechanistic interpretability often identifies circuits inside Transformer models, but explanations of those circuits are usually validated through examples, ablations, and manual reasoning. This leaves a gap between finding a plausible circuit and proving what the circuit does. We introduce Verifiable Transformers, a framework for converting task-localized Transformer circuits into bounded, solver-checkable claims. Given a behavior, a finite task domain, and a candidate-token projection, we extract a task circuit and verify properties such as projected functional equivalence, edge necessity, task-relevant invariance, and final-residual robustness. Direct verification encodes the extracted circuit itself into an SMT solver. When a circuit contains operators that are not exactly or tractably encodable, surrogate-mediated verification fits an SMT-encodable surrogate, validates it against the extracted circuit over the bounded domain, and verifies symbolic explanations against the surrogate. We instantiate direct verification with a GPT-style architecture using Signed L1 BandNorm, sparsemax attention, and LeakyReLU. On small symbolic sequence tasks, we train an SMT-representable Transformer, extract sparse circuits for quote closing and bracket type tracking, and exhaustively verify projected functional equivalence, content invariance, edge necessity, and final-residual robustness. At GPT-2 scale, the same operator stack trains stably on OpenWebText, although naive direct SMT verification remains intractable. We also demonstrate surrogate-mediated verification on task-localized circuits with hard-to-encode attention, showing both verified symbolic explanations and solver-generated counterexamples. The goal is not full-model verification, but a concrete path for turning mechanistic circuit explanations into formal propositions that can be proven or refuted.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses counterexample, robustness.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.24008unread
CAFD: Concept-Aware DNN Fault Detection using VLMs
Amin Abbasishahkoo, Mahboubeh Dadkhah, Lionel Briand · 2026-05-26
arXiv:2605. 24008v1 Announce Type: new Abstract: Fault detection for Deep Neural Networks (DNNs) has received increasing attention in recent years.
Read next because CAFD: Concept-Aware DNN Fault Detection using VLMs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, eval, source, line, rate, full, trained, language. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24008v1 Announce Type: new Abstract: Fault detection for Deep Neural Networks (DNNs) has received increasing attention in recent years. While more advanced hybrid approaches have been proposed to combine multiple sources of information and outperform earlier techniques, they often incur substantial computational overhead, limiting scalability and practicality in real-world settings. In this paper, we introduce Concept-Aware Fault Detection (CAFD), a learning-based approach that achieves superior fault detection performance by effectively integrating multiple information sources while maintaining practical efficiency. Specifically, CAFD is trained using a carefully selected set of informative features, including model-based signals derived from the DNN's outputs, distance-based features, and a novel concept-based feature, called Concept Failure Ratio (CFR). CFR leverages Vision-Language Models (VLMs) to extract textual concepts from images and quantify the likelihood that their presence is associated with DNN failures. By incorporating this feature, CAFD benefits from complementary semantic information, enabling more effective fault detection. Our results demonstrate that CFR serves as an effective indicator for DNN fault detection. We conduct an extensive empirical evaluation of CAFD, comparing it against five state-of-the-art baselines across three subject DNN models and datasets, including ImageNet. Across a wide range of constrained selection budgets, CAFD consistently outperforms all baselines in Fault Detection Rate (FDR), achieving average FDR improvements of 18.3% across all investigated subjects and budget sizes.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, evaluation.
- score 100arxiv cs.LG (Machine Learning)arxiv:2605.23978unread
Algometrics: Forecasting Under Algorithmic Feedback
Marc Schmitt · 2026-05-26
arXiv:2605. 23978v1 Announce Type: new Abstract: In algorithmic markets, predictive models become part of the data-generating process they aim to forecast.
Read next because Algometrics: Forecasting Under Algorithmic Feedback overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, line, control, alone, model. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.23978v1 Announce Type: new Abstract: In algorithmic markets, predictive models become part of the data-generating process they aim to forecast. Once their outputs are converted into trades, allocations, execution schedules, or risk controls, they change the future data on which they are evaluated. I introduce algometrics, a framework for time series whose evolution depends on the predictive algorithms forecasting them. The framework distinguishes historical risk, measured under passive forecasting, from deployment risk, measured when forecasts drive actions. I prove three results. First, deployment risk is not identifiable from passive historical data alone: even in a one-step linear feedback model, infinitely many algorithm-mediated environments induce the same historical law while implying different deployment risks for the same forecaster. Second, historical model rankings can invert under crowding, so a predictor with lower passive error can have higher deployment error once similar algorithms are adopted. Third, randomized or instrumented actions identify short-horizon linear feedback, and I derive a finite-sample bound for deployment-risk estimation. These results suggest that time-series benchmarks in algorithmic markets should report feedback sensitivity alongside predictive accuracy.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24285unread
Memory, Roughness, and Information Persistence in Financial Markets: A Structural Approach to Volatility Forecasting
Akash Deep, Nicholas Appiah, Svetlozar T. Rachev · 2026-05-26
arXiv:2605. 24285v1 Announce Type: cross Abstract: This paper studies the joint role of long-memory dynamics,rough-volatility behavior, and persistence-based forecasting features in equity volatility modeling.
Read next because Memory, Roughness, and Information Persistence in Financial Markets: A Structural Approach to Volatility Forecasting overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rate, does, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24285v1 Announce Type: cross Abstract: This paper studies the joint role of long-memory dynamics,rough-volatility behavior, and persistence-based forecasting features in equity volatility modeling. We combine semiparametric long-memory estimation, rough-volatility diagnostics, and structured forecasting regressions to examine whether persistence measures contain economically meaningful forecasting information beyond conventional volatility predictors. Using a panel of 115 S&P500 constituents from November 2001 through April 2026, we document that volatility proxies exhibit substantial long-memory behavior and locally rough dynamics. The cross-sectional mean Geweke-Porter-Hudak estimate of the memory parameter is $\hat{d} = 0.226$, while the corresponding local-Whittle estimate is $\hat{d} = 0.440$, with statistical significance observed across nearly the entire panel. Rolling estimates of persistence rise substantially during the global financial crisis and the COVID period and display a positive contemporaneous association with the VIX. We then examine whether persistence-related features improve out-of-sample volatility forecasts beyond standard HAR and HAR-X benchmarks. Incorporating cross-sectional persistence aggregates, sectoral persistence measures, and persistence-by-stress interaction terms produces moderate but statistically significant forecasting improvements, particularly at longer horizons and during stress regimes. Forecast gains are strongest during periods of elevated market volatility and in volatility-managed portfolio applications. The results suggest that persistence measures may serve as useful reduced-form indicators of the duration and propagation of uncertainty in financial markets, although the paper does not claim structural identification of the economic mechanisms generating persistence.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24274unread
A lift for input-convex neural network training
Ali Siahkoohi, Anirudh Thatipelli · 2026-05-26
arXiv:2605. 24274v1 Announce Type: new Abstract: Input-convex neural networks (ICNNs) are widely used for log-concave density estimation, convex-potential normalizing flows, optimal transport, and transport-map inversion for high-dimensional Bayesian posteriors.
Read next because A lift for input-convex neural network training overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, rect, soft, rate, recipe, project, trained, emit. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24274v1 Announce Type: new Abstract: Input-convex neural networks (ICNNs) are widely used for log-concave density estimation, convex-potential normalizing flows, optimal transport, and transport-map inversion for high-dimensional Bayesian posteriors. These tasks share a structural constraint: the inter-layer weights of the ICNN must remain non-negative. The standard recipe, projected gradient descent (PGD) onto the non-negative cone, applies a hard, non-smooth projection -- the stiff-penalty limit of an ADMM-style constraint splitting -- and its classical convergence guarantees do not transfer to the non-smooth ICNN training landscape; the differentiable alternative, softplus reparametrization, attenuates the gradient exponentially in the weight magnitude, stalling training with dead inter-layer weights and plateaued loss. Inspired by parameter-extension lifts of PDE-constrained inverse problems, we propose the lift: instead of constraining the inter-layer weights directly, we train an unconstrained hypernetwork that emits them from a permutation-invariant summary of the input batch. This adds stochasticity to the training dynamics that softens the loss landscape, letting the iterates escape the gradient-attenuated region where direct softplus stalls. We trace this softening to three structural ingredients -- a learnable bias acting as slack, a hypernetwork body that conditions on the target batch, and a cross-covariance coupling the two through batch stochasticity -- and prove each one necessary: deleting any single ingredient collapses the cross-covariance that carries the softening. On log-concave energy-based modeling from one-dimensional toy targets to image-flavored latents, and convex-potential normalizing flows on a 21-dimensional tabular benchmark, we show that the lift reaches a lower test loss than both PGD and direct softplus, and turns a plateau-bounded training trajectory into a valley-descending one.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses bias, negative, benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24243unread
GIBLy: Improving 3D Semantic Segmentation through an Architecture-Agnostic Lightweight Geometric Inductive Bias Layer
Diogo Lavado, Alessandra Micheletti, Cl\`audia Soares · 2026-05-26
arXiv:2605. 24243v1 Announce Type: cross Abstract: In 3D scene understanding, deep learning models rely on large models and extensive training to capture basic geometric structures that are present in the 3D data.
Read next because GIBLy: Improving 3D Semantic Segmentation through an Architecture-Agnostic Lightweight Geometric Inductive Bias Layer overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation". Matching terms: under, line, rate, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24243v1 Announce Type: cross Abstract: In 3D scene understanding, deep learning models rely on large models and extensive training to capture basic geometric structures that are present in the 3D data. However, existing methods lack explicit mechanisms to incorporate geometric information, such as learnable primitive shapes, often necessitating large models and more training data which in turn increases cost and can limit generalization. We introduce GIBLy, a lightweight geometric inductive bias layer that integrates learnable geometric priors into 3D segmentation pipelines. GIBLy enhances existing architectures -- whether MLP-based, convolution-based, or transformer-based -- by providing features aligned with simple geometric shapes (and thus human-interpretable) that improve segmentation performance with minimal computational overhead. We validate our approach across multiple 3D semantic segmentation benchmarks, demonstrating consistent performance gains, including up to +11.5% mIoU on TS40K with PTV3, while adding only 58K extra parameters. Our results highlight the benefit of explicitly encoding geometric structure to support accurate and efficient 3D scene understanding, with a lightweight add-on layer
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses bias, benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24212unread
Distributionally Robust Transfer Learning with Structurally Missing Covariates, with Application to Cross-National Cardiac Arrest Prediction
Siqi Li, Chuan Hong, Ziye Tian, Benjamin Sieu-Hon Leong, Koshi Nakagawa, Hideharu Tanaka, Sang Do Shin, Khuong Quoc Dai, Do Ngoc Son, Marcus Eng Hock Ong, Nan Liu, Molei Liu · 2026-05-26
arXiv:2605. 24212v1 Announce Type: cross Abstract: Deploying clinical prediction models across healthcare systems often fails when key training covariates are unavailable at deployment and labeled outcomes are limited in the target domain.
Read next because Distributionally Robust Transfer Learning with Structurally Missing Covariates, with Application to Cross-National Cardiac Arrest Prediction overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: class, rect, under, correct, distributional, source, line, rate. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24212v1 Announce Type: cross Abstract: Deploying clinical prediction models across healthcare systems often fails when key training covariates are unavailable at deployment and labeled outcomes are limited in the target domain. For example, high-performing models for out-of-hospital cardiac arrest (OHCA) rely on detailed prehospital measurements routinely collected in high-resource settings but unavailable in many international registries. Existing methods either discard missing covariates, sacrificing predictive information, or rely on untestable assumptions about their target distribution. We propose DRUM (\underline{D}istributionally \underline{R}obust \underline{U}nsupervised transfer learning with structurally \underline{M}issing covariates), a framework that transfers prediction models to target populations where certain covariates are structurally absent and outcome labels are unavailable. DRUM partitions covariates into shared components ($X$), observed across all settings, and missing components ($A$), observed only in the source. Rather than imputing missing covariates, DRUM optimizes worst-case predictive performance over the unknown target distribution of $A \mid X$ using a neural network generator, with a robustness parameter controlling allowable deviation from the source conditional. We further develop a bias correction procedure that reduces sensitivity to nuisance estimation error. Simulations show substantial improvements in both mean and worst-case prediction error under distribution shift. Applied to cross-national OHCA prediction, transferring models from a US registry to multiple Asian registries where prehospital variables are unrecorded, DRUM yields better-calibrated predictions and improved clinical classification performance across sites.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses bias, robustness.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24210unread
Characterizing the Representational Capacity of Neural Processes
Robin Young · 2026-05-26
arXiv:2605. 24210v1 Announce Type: new Abstract: What functions can Neural Processes represent?
Read next because Characterizing the Representational Capacity of Neural Processes overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rate. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24210v1 Announce Type: new Abstract: What functions can Neural Processes represent? We analyze the representational capacity of popular NP architectures: Conditional Neural Processes (CNPs), Attentive Neural Processes (ANPs), Transformer Neural Processes (TNPs), and their latent variants. We prove these architectures form a strict hierarchy. CNP-representable functions are exactly those depending on finitely many expected features of the context distribution. ANPs strictly generalize CNPs via query-dependent reweighting, enabling kernel smoothers. ConvCNPs and ANPs are incomparable; each contains functions outside the other, separated by stationarity versus translation equivariance. TNPs with $L$ self-attention layers capture $L$-hop context interactions. For latent NPs, we show finite-dimensional latents provide coherent sampling but do not circumvent encoder limitations; matching GP posterior distributions requires latent dimension scaling with context size. These results provide a theoretical foundation for architecture selection based on task structure.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses limitation, limitations.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.26087unread
DiscoverPhysics: Benchmarking LLMs for Out-of-the-Box Scientific Thinking
Matt L. Wiemann, Lindsay M. Smith, Peter Melchior, Siddharth Mishra-Sharma, Andrew Gordon Wilson, Pavel Izmailov, Carolina Cuesta-L\'azaro · 2026-05-26
arXiv:2605. 26087v1 Announce Type: new Abstract: Frontier LLMs now perform strongly across a wide range of physics evaluations, but it is hard to disentangle genuine reasoning from recall of established science.
Read next because DiscoverPhysics: Benchmarking LLMs for Out-of-the-Box Scientific Thinking overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, good, eval, source, rate, implement, does. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.26087v1 Announce Type: new Abstract: Frontier LLMs now perform strongly across a wide range of physics evaluations, but it is hard to disentangle genuine reasoning from recall of established science. We introduce DiscoverPhysics, an interactive benchmark that asks a LLM agent to discover the laws of motion of a simulated world whose physics deliberately deviates from our own. We construct 22 worlds governed by, among others, screened and fractional-power gravity, multi-species couplings, hidden dark-matter-like particles, non-coordinate-free physics, and time-varying interactions. Each world is generated on demand by an N-body simulator, for which the agent proposes several rounds of experiments, observes raw trajectory data, and ultimately submits both a natural-language explanation of the world's physics and a Python implementation of the inferred law. Because solving a world requires the agent to design informative experiments and revise its hypotheses, the benchmark probes long-horizon reasoning over an experimental history. We evaluate submissions along two complementary axes: trajectory MSE on held-out particles and an LLM-judged explanation score following an expert-written rubric assessing conceptual understanding of each world. Across eleven frontier models, we find that the strongest agents pass only half of the worlds and consistently fail on those where latent structure must be uncovered. Open-source models lag substantially behind commercial models, both in their ability to design informative experiments and in extracting conclusions from the data. We further find that good predictive accuracy does not guarantee high explanation quality and that conceptual understanding depends on hypothesis refinement through well-chosen experiments.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25773unread
Efficient Benchmarking Is Just Feature Selection and Multiple Regression
Sam Bowyer, Acyr Locatelli, Kris Cao · 2026-05-26
arXiv:2605. 25773v1 Announce Type: new Abstract: Efficient benchmarking techniques aim to lower the computational cost of evaluating LLMs by predicting full benchmark scores using only a subset of a benchmark's questions.
Read next because Efficient Benchmarking Is Just Feature Selection and Multiple Regression overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, under, eval, full, stage, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25773v1 Announce Type: new Abstract: Efficient benchmarking techniques aim to lower the computational cost of evaluating LLMs by predicting full benchmark scores using only a subset of a benchmark's questions. By reframing this problem as an instance of multiple regression with feature selection, we find that existing efficient benchmarking methods can be greatly improved by simply using kernel ridge regression at the prediction stage. Additionally, using an information-theoretic feature-selection algorithm called minimum redundancy maximum relevance (mRMR), we can further improve upon these methods by selecting question subsets that will be maximally useful for prediction. Except in very data-poor settings, these approaches consistently achieve smaller prediction errors (in both MAE and RMSE), and greater ranking correlation between predicted and true scores (in both Spearman $\rho$ and Kendall $\tau$) across a range of benchmarks using both binary and continuous metrics. Furthermore, mRMR subsampling is much faster than competitor methods (which often involve fitting probabilistic models or running clustering algorithms), and is more likely to select the same questions under different random seeds or training data splits. Tutorial code can be found at https://github.com/sambowyer/mrmr_eval .
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25648unread
StrTransformer: Source-Wise Structured Transformers for Unsupervised Blind Source Recovery
Yuan-Hao Wei · 2026-05-26
arXiv:2605. 25648v1 Announce Type: new Abstract: This paper proposes StrTransformer, a source-wise structured Transformer framework for blind source recovery and branch-wise latent modeling.
Read next because StrTransformer: Source-Wise Structured Transformers for Unsupervised Blind Source Recovery overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rect, under, eval, source, token, control, symmetry. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25648v1 Announce Type: new Abstract: This paper proposes StrTransformer, a source-wise structured Transformer framework for blind source recovery and branch-wise latent modeling. Instead of using an encoder to infer latent variables, StrTransformer directly optimizes the latent source matrix together with an observation-space mixer and source-wise structural Transformer branches. The mixer enforces reconstruction consistency, while each Transformer branch imposes a differentiable structural constraint on one latent source trajectory. Specifically, each source is converted into multi-scale patch tokens, randomly masked, processed by a locality-biased Transformer, and evaluated through a masked patch reconstruction energy. This energy acts as an implicit source-wise structural prior. To encourage different latent branches to specialize into different temporal regimes, StrTransformer further introduces an ordered multi-scale controller that learns branch-specific patch-scale weights, ordered scale centers, and locality attention slopes. The resulting objective combines observation reconstruction, source-wise structural regularization, and modular auxiliary penalties for separation and scale specialization. We analyze the decoupling and coupling structure of the objective, the regularized exact-reconstruction fiber, and the reduction of permutation symmetry induced by ordered branch descriptors. A controlled case study shows that the learned branches converge to distinct temporal-scale structures and recover source-aligned latent trajectories under post-hoc evaluation.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, evaluation.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.25290unread
Choosing Online Experiment Designs under Interference in Ads, Recommendations, and Member-Experience Systems
Prashant Shekhar, Caroline Howard · 2026-05-26
arXiv:2605. 25290v1 Announce Type: new Abstract: Online experiments in ads, recommendation, and member-experience systems are often planned before the dominant interference mechanism is known.
Read next because Choosing Online Experiment Designs under Interference in Ads, Recommendations, and Member-Experience Systems overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe". Matching terms: under, line, implement, compare, control, propagate. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25290v1 Announce Type: new Abstract: Online experiments in ads, recommendation, and member-experience systems are often planned before the dominant interference mechanism is known. A treatment may propagate through budgets, inventory, producer exposure, graph spillovers, or temporal carryover, making the randomization design itself a statistical decision. We formulate this problem as robust design selection over uncertain exposure mechanisms. Given a finite catalog of six implementable designs, the selector compares each design by worst-case planning risk over an ambiguity set. The risk combines exposure bias, assignment-unit variance, minimum detectable effect, contamination or carryover, operational cost, and estimand mismatch. For theoretical justification, the paper develops a geometry-aware guarantee, stating that design bias is bounded by Wasserstein distance to the launch exposure distribution, and this penalty is minimax tight under Lipschitz exposure response. We also prove finite-catalog approximation and a robust selector theorem with excess-risk control, exact recovery under separation, and certified shortlists when the risk surface is flat. Empirically, the same selector gives different recommendations across samples from public datasets. It selects user-randomization on Criteo ads with dimensionless robust risk 1.295, switchbacks on Open Bandit-bts/men with risk 2.105, and cluster-randomization on KuaiRand with risk 2.240. The Open Bandit case stresses known but uneven logging support, with propensities from 0.00006 to 0.594 and a 5.17% IPS effective-sample share. Overall, the paper contributes an interference-aware experiment design framework based on mechanism-robust design decisions, where the output is either a justified design choice or an uncertainty shortlist.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses bias.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24422unread
Clustering based on Stochastic Dominance with application for risk averters and risk seekers
Hua Li, Xue Jia, Yilin Kang, Wing-Keung Wong · 2026-05-26
arXiv:2605. 24422v1 Announce Type: new Abstract: Stochastic Dominance (SD) theory provides a rigorous framework for selecting superior assets tailored to the asset allocation needs of investors with varying risk preferences (i.
Read next because Clustering based on Stochastic Dominance with application for risk averters and risk seekers overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, eval, rate, test, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24422v1 Announce Type: new Abstract: Stochastic Dominance (SD) theory provides a rigorous framework for selecting superior assets tailored to the asset allocation needs of investors with varying risk preferences (i.e., risk-averse, risk-seeking, and risk-neutral). However, traditional stock clustering methods typically rely on geometric metrics such as Euclidean distance, which often fail to effectively capture the intrinsic risk dominance relationships among assets. To address this limitation, this paper proposes an innovative clustering analysis framework based on SD test statistics. Methodologically, this study deeply integrates SD theory with machine learning algorithms. Transcending the limitations of traditional reliance on geometric distance, we innovatively utilize test statistics from first-, second-, and third-order SD to construct a "Stochastic Dominance Coefficient Matrix." Building upon this matrix, we modify the classic K-means and Hierarchical Clustering algorithms. Specifically, we derive 12 distinct algorithm variants tailored to different orders of SD relationships. Simultaneously, we construct the SD-SC coefficient and the SD-DBI index as specialized validity indices to evaluate the clustering performance. Empirically, we analyze constituent stock data from a representative developed market (the US NASDAQ Index) and an emerging market (China's CSI 100 Index). The results verify the effectiveness and robustness of the proposed method. Furthermore, we apply the clustering results to the modification of the Single Index Model and the construction of Global Minimum Variance Portfolios (GMVP). The findings demonstrate that the proposed method effectively facilitates customized asset allocation for investors, holding significant theoretical value and practical implications.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses limitation, limitations, robustness.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24364unread
Multicalibration Boosting: Theory, Convergence, and Transferability
Hanxuan Ye, Hongzhe Li · 2026-05-26
arXiv:2605. 24364v1 Announce Type: new Abstract: Multicalibration extends classical calibration by requiring predictions to be unbiased over a rich collection of functions, encompassing both prediction slices and subpopulations.
Read next because Multicalibration Boosting: Theory, Convergence, and Transferability overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: class, under, rate, project, control, position, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24364v1 Announce Type: new Abstract: Multicalibration extends classical calibration by requiring predictions to be unbiased over a rich collection of functions, encompassing both prediction slices and subpopulations. It has emerged as a powerful framework for fairness, robustness, and reliable prediction, yet the theoretical understanding of multicalibration boosting (MCBoost) remains fragmented and often relies on restrictive assumptions. In this work, we develop a unified and refined perspective on MCBoost that subsumes existing variants, including multiaccuracy, BatchGCP, and BatchMVP. We uncover several phenomena that provide new insights into its practical behavior: even highly accurate and flexible predictors can remain substantially miscalibrated; enforcing multicalibration introduces a calibration-risk trade-off; and early stopping plays a central role in controlling this trade-off. On the theoretical side, we establish a general framework for MCBoost under weaker and more realistic conditions. We show that the boosting iterates converge to a Bregman projection of the population-optimal predictor onto the cumulative span generated by the audit class, thereby explicitly characterizing the function space on which multicalibration is achieved. We further derive convergence rates under different smoothness assumptions, finite-sample guarantees, and principled stopping rules that ensure multicalibration at termination. Finally, we extend the theory of universal adaptability under covariate shift, providing more general transfer guarantees and clarifying when multicalibrated predictors generalize across domains. These results provide a more complete theoretical foundation and practical guidance for multicalibration boosting, positioning it as both a unifying framework and a reliable post-processing approach for modern predictive models.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses bias, robustness.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24244unread
MEDAL: Manifold Embedding Distillation via Autoencoder Learning
Irene Chang, Tarek M. Zikry, Genevera I. Allen · 2026-05-26
arXiv:2605. 24244v1 Announce Type: new Abstract: Low-dimensional embeddings are widely used as visual summaries of high-dimensional data and to enable downstream scientific discoveries.
Read next because MEDAL: Manifold Embedding Distillation via Autoencoder Learning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, eval, line, without, alone, trained, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24244v1 Announce Type: new Abstract: Low-dimensional embeddings are widely used as visual summaries of high-dimensional data and to enable downstream scientific discoveries. Yet, popular nonlinear dimension reduction methods, such as t-SNE and UMAP, are often selected based on visual appeal alone and without rigorous quantitative validation. A major reason is that manifold embeddings typically do not provide an out-of-sample map nor an inverse back to the original feature space; this makes held-out validation, the gold standard in supervised learning, all but impossible. To address these challenges, we develop a novel framework, MEDAL (Manifold Embedding Distillation via Autoencoder Learning), which distills a fitted manifold embedding into a reusable encoder--decoder model. MEDAL trains a constrained autoencoder whose bottleneck exactly matches any teacher embedding while the decoder reconstructs the original input; this yields an explicit map for new samples, an approximate inverse, and a pointwise reconstruction-based measure of distortion in the manifold space. This converts static manifold embeddings into models that can be evaluated on held-out data, enabling quantitative validation including comparing different dimension reduction methods as well as hyperparameter tuning. Across multiple benchmark and scientific case studies, we show that MEDAL enables held-out validation to determine optimal manifold embeddings and hyperparameters, reveals biologically coherent regions that are difficult to preserve in two dimensional embeddings, and detects distribution shift when new samples are mapped into a fixed reference manifold. MEDAL provides a general validation wrapper to any existing dimension reduction technique that will improve the rigor and
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv stat.ML (Machine Learning)arxiv:2605.24076unread
Causality as the Statistical Conscience of Artificial Intelligence: From Pearl's Ladder to Trustworthy Machines
Ernest Fokou\'e · 2026-05-26
arXiv:2605. 24076v1 Announce Type: new Abstract: Modern Artificial Intelligence achieves remarkable predictive power by optimizing statistical risk functionals over vast corpora.
Read next because Causality as the Statistical Conscience of Artificial Intelligence: From Pearl's Ladder to Trustworthy Machines overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, under, rate, without, language, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.24076v1 Announce Type: new Abstract: Modern Artificial Intelligence achieves remarkable predictive power by optimizing statistical risk functionals over vast corpora. Yet a gap separates this from genuine intelligence: the inability to distinguish correlation from causation. This paper argues that causal inference (identifying mechanisms invariant under intervention) is AI's indispensable statistical conscience. Without causal grounding, AI systems are correlation machines: powerful in familiar domains, brittle under distribution shift, and biased in high-stakes settings. Three contributions develop this argument. First, a Statistical Necessity Theorem for Causal Generalization: any algorithm achieving out-of-distribution generalization must encode causal structure, formalizing the distinction between prediction P(Y|X) and intelligence P(Y|do(X)). Second, a unified framework connects Pearl's do-calculus, the Potential Outcomes framework, Double Machine Learning, and Invariant Risk Minimization as a family of Causal Statistical Estimators, each identifying interventional distributions under different assumptions. Third, three AI failure modes (hallucination in large language models, reward hacking in reinforcement learning from human feedback, and degradation under distribution shift) are manifestations of causal blindness, each admitting a principled statistical remedy. Trustworthy AI is, at its core, a problem of causal statistics. The statistical community is not merely equipped to solve it -- it is the only community with the foundational tools to do so rigorously.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, bias.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24941unread
Memory-Induced Tool-Drift in LLM Agents
Mahavir Dabas, Jihyun Jeong, Ming Jin, Ruoxi Jia · 2026-05-26
arXiv:2605. 24941v1 Announce Type: new Abstract: Modern LLM agents combine long-term memory for personalization with tool-calling interfaces for taking actions in the world -- a combination underpinning contemporary production systems.
Read next because Memory-Induced Tool-Drift in LLM Agents overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, persona, word, rect, under, line, rate, contexts. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24941v1 Announce Type: new Abstract: Modern LLM agents combine long-term memory for personalization with tool-calling interfaces for taking actions in the world -- a combination underpinning contemporary production systems. We study a previously unexamined failure of this combination: when personality-driven biases stored in memory (cost-consciousness, impatience, risk tolerance, etc.) silently affect tool calls in contexts where they are not applicable. We call this memory-induced tool-drift and operationalize it through MEMDRIFT, a benchmark of 105 scenarios spanning five bias dimensions and seven professional domains, generated through an automated adversarial pipeline. Across seven frontier models -- including those with extended reasoning -- biased memories raise deflection scores (a judge-scored measure of parameter deviation from unbiased baselines) by up to $+3.6$ points on a 1--5 scale. Tool-drift persists when memory management is handled by three production memory architectures. The phenomenon affects real-world tools: scanning 6{,}062 tools across 288 verified MCP servers, we flag 608 with susceptible parameters and confirm tool-drift on a validated subset. Mechanistically, biased memories act as implicit steering vectors, pushing activations along the same latent directions as explicit behavioral instructions. They also redistribute attention from task-relevant context toward memory entries with surface-level keyword overlap to the target parameter. Standard defenses -- prompt-based relevance instructions and memory filters -- reduce drift but do not eliminate it. As agents take increasingly consequential actions on a user's behalf, memory-induced tool-drift represents a systematic vulnerability that current safeguards do not address, motivating dedicated defenses at the intersection of memory management and tool-call generation.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, bias, adversarial, benchmark.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24903unread
SEED: Semi-supervised Continual MalwarE Detection for Tackling ConcEpt Drift on a BuDget
Suresh Kumar Amalapuram, Bikraj Shresta, Siva Ram murthy Chebiyam, Bheemarjuna Reddy Tamma, Sumohana S Channappayya · 2026-05-26
arXiv:2605. 24903v1 Announce Type: new Abstract: Machine learning based malware detectors become obsolete over time due to concept drift in benign and malware applications.
Read next because SEED: Semi-supervised Continual MalwarE Detection for Tackling ConcEpt Drift on a BuDget overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, under, eval, rate, compare, project, full, position. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24903v1 Announce Type: new Abstract: Machine learning based malware detectors become obsolete over time due to concept drift in benign and malware applications. Recent methods rely on fully labeled data and use hierarchical contrastive loss (HCL) with active learning to improve robustness against drift by exploiting semantic structure in malware representations. However, obtaining labeled data in the security domain is difficult. Under partially labeled settings, HCL suffers significant performance degradation in detecting unseen malware, especially on datasets such as BODMAS where strong semantic structure may not exist. In this paper, we propose SEED, a semantic-structure-agnostic method for malware detection under limited supervision. SEED combines a tailored binary cross-entropy objective with semi-supervised continual learning and active learning. For partially labeled seen tasks, unlabeled samples are projected into a representation space constructed from previously seen data using singular value decomposition, and paired with suitable labeled samples to encourage representation consistency. For unseen tasks with fully unlabeled data, uncertainty is quantified using cosine distance in representation space, and the most uncertain samples are selected for analyst labeling. We evaluate SEED on both Windows and Android malware datasets. Using only 20% labeled data on seen tasks, SEED achieves average AUT improvements of 40% on BODMAS and 14% on AndroZoo for unseen malware detection compared to HCL* (the semi-supervised adaptation of HCL), while remaining competitive on APIGraph. Finally, we introduce a delayed buffer update strategy to reduce label noise propagation during replay and improve learning stability.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24834unread
Reflect-Guard: Enhancing LLM Safeguards against Adversarial Prompts via Logical Self-Reflection
Lixing Lin, Juli You, Yue Li, Luyun Lin, Yiqing Wang, Zhen Zhang, Moxuan Zheng · 2026-05-26
arXiv:2605. 24834v1 Announce Type: new Abstract: Large language model (LLM) safety classifiers such as Llama Guard are effective at detecting overtly harmful prompts but remain vulnerable to adversarial jailbreak attacks that disguise malicious intent through role-play scenarios, fictional framing, and indirect requests.
Read next because Reflect-Guard: Enhancing LLM Safeguards against Adversarial Prompts via Logical Self-Reflection overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)". Matching terms: class, rect, rate, chain, test, lora, language, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24834v1 Announce Type: new Abstract: Large language model (LLM) safety classifiers such as Llama Guard are effective at detecting overtly harmful prompts but remain vulnerable to adversarial jailbreak attacks that disguise malicious intent through role-play scenarios, fictional framing, and indirect requests. We present Reflect-Guard, a method that augments LLM-based safety classifiers with chain-of-thought self-reflection capabilities through parameter-efficient fine-tuning. Our approach distills analytical reasoning from GPT-4o-mini into structured reflection annotations, then trains Llama-Guard-3-8B via QLoRA to generate logical self-reflections before issuing safety verdicts. Using only 1000 training examples and updating just 0.5% of model parameters (~42M), Reflect-Guard achieves substantial improvements on two challenging benchmarks. On WildGuardTest, F1 score improves from 0.770 to 0.842 (+7.2 pp), with recall on adversarial prompts increasing from 0.513 to 0.921 (+40.8 pp). On JailbreakBench, the attack success rate drops from 10.3% to 1.8%, representing an 82.5% relative reduction. These gains are especially pronounced on adversarial inputs, where the explicit reasoning step enables the model to see through obfuscation techniques that defeat standard pattern-matching approaches. Our results demonstrate that teaching safety classifiers to reason about adversarial intent, rather than simply classify surface patterns, is a promising direction for robust LLM safety.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses adversarial, benchmark.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24817unread
RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry
Bo Lv, Zhiheng Xu, KeDong Xiu, Ruyi Ding, Tianhang Zheng, Zhibo Wang, Kui Ren · 2026-05-26
arXiv:2605. 24817v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures have become an increasingly important paradigm for scaling Large Language Models (LLMs).
Read next because RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, fill, under, eval, source, line, rate, test. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24817v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures have become an increasingly important paradigm for scaling Large Language Models (LLMs). As MoE models are increasingly deployed in real-world services, safety auditing becomes necessary to verify whether these models produce or facilitate harmful behaviors during operation. However, existing content-based auditing methods typically require access to user prompts, model inputs, or generated outputs, potentially exposing sensitive user information and creating a fundamental tension between LLM safety and user privacy. On the other hand, we observe that, in MoE models, sparse expert routing maps different inputs to activate different expert-execution patterns, producing measurable footprints in low-level GPU execution telemetry. Inspired by this observation, we propose RouteScan, a non-intrusive auditing framework for detecting harmful behaviors through GPU-level expert routing telemetry. Specifically, RouteScan utilizes the number of active GPU threads allocated to expert modules during the prefilling phase as a discriminative micro-architectural fingerprint, and builds a lightweight detection pipeline that isolates cross-domain invariant risk indicators for the precise identification of malicious prompts. Comprehensive evaluations on open-source MoE LLMs with distinct routing designs demonstrate that RouteScan achieves strong generalization, with an AUROC exceeding 0.93 on unseen harmful domains and 0.96 under novel jailbreak wrappers. Moreover, empirical inversion tests show that the collected expert routing telemetry provides limited information for prompt reconstruction, suggesting a practical privacy advantage over content-based auditing methods.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24765unread
CyberMaskQA: A Privacy-Aware Benchmark for Evaluating Large Language Models in Cybersecurity Question Answering
Matilda Gaddi, Jin Noh, Onat Gungor, Tajana Rosing · 2026-05-26
arXiv:2605. 24765v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly applied to cybersecurity question answering (QA) for critical tasks such as incident response and vulnerability analysis.
Read next because CyberMaskQA: A Privacy-Aware Benchmark for Evaluating Large Language Models in Cybersecurity Question Answering overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, eval, line, rate, control, contexts, test, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24765v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly applied to cybersecurity question answering (QA) for critical tasks such as incident response and vulnerability analysis. However, real-world operational contexts, including system logs and network configurations, inherently contain sensitive identifiers, e.g., IP addresses, host names, and user accounts. Processing this data with cloud-based models is often unsafe or infeasible in regulated environments. Furthermore, progress in privacy-preserving QA is hindered by the lack of annotated, context-rich datasets capable of jointly evaluating operational reasoning and privacy preservation. To address this gap, we introduce CYBERMASKQA, a privacy-aware QA benchmark covering key security domains. Unlike existing benchmarks that primarily test factual knowledge, CYBERMASKQA grounds questions in realistic organizational contexts with explicit causal dependencies among assets and privileges. Generated through a systematic pipeline, the dataset combines human-curated base scenarios with LLM-driven semantic expansion, annotating each instance with precise private entity labels to enable controlled information disclosure. Evaluations of QA accuracy and masking performance demonstrate the benchmark's utility for developing deployable, context-aware cybersecurity models and facilitating nuanced studies of privacy-utility trade-offs. Upon acceptance, we will release the dataset and the generation framework.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24696unread
CALIBURN: A Regime-Sensitivity Study of Operationally Calibrated Streaming Intrusion Detection
Michel A. Youssef · 2026-05-26
arXiv:2605. 24696v1 Announce Type: new Abstract: Streaming network intrusion detection systems must process flows continuously while keeping memory bounded, but most current methods leave alerting threshold selection as a post-hoc tuning problem poorly suited to production.
Read next because CALIBURN: A Regime-Sensitivity Study of Operationally Calibrated Streaming Intrusion Detection overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, class, under, alpha, eval, line, rate, control. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24696v1 Announce Type: new Abstract: Streaming network intrusion detection systems must process flows continuously while keeping memory bounded, but most current methods leave alerting threshold selection as a post-hoc tuning problem poorly suited to production. Operators need alerting behaviour specifiable before deployment using inputs such as false-negative cost, false-positive cost, and alerting budget. This paper presents CALIBURN, a five-component streaming alerting pipeline composed of a truncated Bayesian online change-point detector, an isotonic calibration layer mapping the change-point posterior to an empirical conditional attack probability, a cost-sensitive decision threshold derived from operator-specified misclassification costs, a Conformal Risk Control wrapper that converts an alert-budget specification into a within-window valid threshold under exchangeability, and a multi-window burn-rate alerting layer adapted from Site Reliability Engineering practice. Rather than claiming uniform dominance, we present CALIBURN as a regime-sensitivity study, evaluating the pipeline across three attack-prevalence regimes: LITNET-2020 at 5.2 percent, CICIDS2017 at 22.06 percent, and UNSW-NB15 at 64 percent. In the rare-attack regime, CALIBURN achieves AUC-PR 0.943 on LITNET-2020, outperforming the best streaming baseline by 2.21x and the best batch reference by 4.12x; isotonic calibration reduces Brier score by 30 percent. In the moderate-prevalence regime, CALIBURN remains the strongest streaming method on CICIDS2017 but is exceeded by batch density methods. In the high-prevalence regime, all streaming methods approach the prevalence floor. We further identify two distinct CRC-collapse mechanisms driving the alert rule to degeneracy at small alpha, treating both as operational guidance for practitioners.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses negative.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24552unread
Ellipsoid Control: A White-list Jailbreak Defense via Benign Latent Modeling
Luoyu Chen, Weiqi Wang, Zhiyi Tian, Feng Wu, Ahmed Asiri, Shui Yu · 2026-05-26
arXiv:2605. 24552v1 Announce Type: new Abstract: Representation engineering (RepE) defenses have shown strong robustness against jailbreak attacks on large language models (LLMs).
Read next because Ellipsoid Control: A White-list Jailbreak Defense via Benign Latent Modeling overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, eval, project, control, without, test, language, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24552v1 Announce Type: new Abstract: Representation engineering (RepE) defenses have shown strong robustness against jailbreak attacks on large language models (LLMs). However, these methods fundamentally rely on black-list supervision: they learn jailbreak-to-refusal activation transformations from harmful or jailbreak data that are inherently incomplete and continuously evolving. Hence, the performance of RepE-based defenses becomes tightly coupled to the quality and coverage of collected harmful samples, leaving models vulnerable to unseen attacks. This reliance also obscures the distinction between defenses that fit known harmful distributions and defenses that protect a benign latent region without estimating the harmful distribution. We adopt the opposite, the white-list perspective, by leveraging the accessibility and abundance of benign data. The goal is to elicit refusal on arbitrary inputs while ensuring that harmless inputs are not falsely rejected. This shifts the core research question to: How can we design a robust benign-latent preservation mechanism such that the benign latent distribution remains intact while refusal is elicited? To answer this, we propose Ellipsoid Control, a test-time defense. It performs projected gradient descent that can elicit refusal on arbitrary inputs, aiming to improve defense effectiveness. At the same time, an anisotropic benign-geometry ellipsoid is fitted from abundant benign data to constrain the update to minimize distortion of the benign latent geometry. This tight constraint helps preserve model utility. Across multiple LLMs, jailbreak attacks, benign tasks, and safety-boundary evaluations, Ellipsoid Control consistently enhances safety while better preserving utility, demonstrating the effectiveness of the white-list approach for jailbreak defense
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses robustness, evaluation.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24551unread
Routing Cybersecurity Awareness Training by FFM Personality Trait: A Quasi-Experimental Evaluation
Glory Okwata, Mohammad A. Razzaque · 2026-05-26
arXiv:2605. 24551v1 Announce Type: new Abstract: Cybersecurity awareness training has historically adopted a one-size-fits-all approach, despite established individual differences in how users process and retain security information.
Read next because Routing Cybersecurity Awareness Training by FFM Personality Trait: A Quasi-Experimental Evaluation overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: persona, eval, line, rate, implement, factor, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24551v1 Announce Type: new Abstract: Cybersecurity awareness training has historically adopted a one-size-fits-all approach, despite established individual differences in how users process and retain security information. Personality has been proposed as one axis along which training content might be tailored; yet no prior study has implemented and empirically evaluated a complete personality-conditional system end-to-end. This paper reports the design, implementation, and quasi-experimental evaluation of \emph{TailoredSec}, a mobile cybersecurity awareness application that routes training content based on a user's dominant Five-Factor Model (FFM) personality trait, as measured by the ten-item Big Five Inventory (BFI-10). Seventy-four UK-based adults were allocated to a traditional video-training condition ($n = 40$) or a personality-conditional condition ($n = 34$). Both groups completed a four-item scenario-based pre-assessment (scored 0--40), a single training session, and an equivalent post-assessment. The personality-conditional group additionally completed the BFI-10 (Big Five Inventory-10) and was routed to one of four training modules covering five FFM traits (Conscientiousness and Neuroticism share a module). Pre-assessment scores did not differ between groups ($t(69.1) = 0.43$, $p = .67$), confirming baseline equivalence. The personality-conditional group scored significantly higher on the post-assessment ($M = 35.88$, $SD = 5.00$ vs $M = 30.75$, $SD = 10.23$; Welch's $t(58.5) = 2.81$, $p = .007$; Cohen's $d = 0.62$; 95\% CI $[1.47, 8.79]$ marks), with a pass-rate of 100\% versus 77.5\% (Fisher's exact $p < .01$). These results offer preliminary support for personality-conditional content routing as a feasible design principle for cybersecurity awareness training.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24535unread
Steering Beyond the Support: Adversarial Training on Unsupervised Jailbroken Activation Simulation
Luoyu Chen, Weiqi Wang, Zhiyi Tian, Chenhan Zhang, Feng Wu, Jianhuan Huang, Ahmed Asiri, Shui Yu · 2026-05-26
arXiv:2605. 24535v1 Announce Type: new Abstract: Jailbreak prompts can trigger harmful completions on aligned LLMs, In accordance, safety steering has been proposed: test-time activation interventions that steer jailbreak activations to trigger refusal while preserving benign utility.
Read next because Steering Beyond the Support: Adversarial Training on Unsupervised Jailbroken Activation Simulation overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, class, latin, completions, rect, rate, test, completion. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24535v1 Announce Type: new Abstract: Jailbreak prompts can trigger harmful completions on aligned LLMs, In accordance, safety steering has been proposed: test-time activation interventions that steer jailbreak activations to trigger refusal while preserving benign utility. However, existing steering methods are fundamentally supervised and tied to a static, limited training set, whereas real jailbreaks evolve and are often out-of-distributed from the training set, leading to failures on unseen attacks. In this paper, we tackle the failure on unseen jailbreaks problem, base on unsupervised latent direction discovery. We propose a bi-level adversarial training framework for zero-shot jailbreak defense. In the inner step, we simulate diverse jail-broken activations by extrapolating from refusal-state harmful-request activations via unsupervised latent direction discovery, which expands the coverage of real jailbreak activation subspaces. In the outer step, we train a potential-induced steering field to push these adversarial jailbroken states into refusal regions while keeping benign unchanged. Across three LLMs and six classical jailbreak families, our method achieves strong defense with attack success rates mostly below 5%, and rising subspace coverage throughout training helps explain the improved generalization.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, adversarial.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24421unread
Poisoning the Watchtower: Prompt Injection Attacks Against LLM-Augmented Security Operations Through Adversarial Log Content
Rohan Pandey, Archit Bhujang · 2026-05-26
arXiv:2605. 24421v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as analyst assistants in security operations centers (SOCs), where they ingest log and alert data to produce triage labels, incident summaries, or remediation advice.
Read next because Poisoning the Watchtower: Prompt Injection Attacks Against LLM-Augmented Security Operations Through Adversarial Log Content overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, persona, class, rect, under, eval, assistant. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24421v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as analyst assistants in security operations centers (SOCs), where they ingest log and alert data to produce triage labels, incident summaries, or remediation advice. We study a structural failure mode of this design: many log fields are attacker controlled. User agents, URLs, payloads, DNS queries, and attempted usernames can therefore carry instructions to the model alongside evidence of the intrusion. We call this setting \emph{log-substrate prompt injection}. We introduce a four-class taxonomy of log-substrate attacks: direct override (S1), persona hijack (S2), context manipulation (S3), and obfuscated payloads (S4). We evaluate 48 strategy-defense-task combinations using \texttt{gpt-4o-mini} as the analyst. Three findings stand out. First, direct overrides are ineffective in our setting: all S1 classification attacks achieve 0\% suppression. In contrast, persona hijacks suppress 68\% of malicious logs under a naive classifier and remain effective under stronger defenses. Second, summarization is the highest-risk task: context manipulation reaches 96\% injection success without defenses and 38\% even with constrained output. Third, defenses reduce but do not eliminate the attack surface: average injection success falls from 26.6\% under naive prompting to 11.8\% under our strongest defense. We also compare empirical results to a deterministic mock analyst and find that simulation substantially mispredicts current model behavior, especially for direct overrides. These results suggest that SOC copilots should treat raw log content as adversarial input rather than ordinary analyst context.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, adversarial.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24309unread
Reframing LLM Agent Security as an Agent-Human Interaction Problem
Peiran Wang, Ying Li, Yuan Tian · 2026-05-26
arXiv:2605. 24309v1 Announce Type: new Abstract: We argue that LLM agent security is fundamentally an agent-human interaction (AHI) problem, not a purely algorithmic one.
Read next because Reframing LLM Agent Security as an Agent-Human Interaction Problem overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Coupling evil personas with wrong answers fails to protect Qwen2.5-7B from EM-induced alignment collapse — and the apparent capability ordering across coupling conditions is mostly eval contamination (LOW confidence)". Matching terms: class, rect, alignment, eval, control, factor, position. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24309v1 Announce Type: new Abstract: We argue that LLM agent security is fundamentally an agent-human interaction (AHI) problem, not a purely algorithmic one. To substantiate this position, we conduct a systematic analysis of 59 academic papers, 21 production agent systems, and 26 security plugins as of April 2026. Our analysis reveals a striking pattern: the three widely deployed human-centric security mechanisms (policy specification, runtime approval, and scope configuration) dominate industry practice, each adopted by at least 14 of 21 systems (14, 15, and 16, respectively), while the categories most heavily studied in academia (intent anchoring and trust labeling) see zero production deployment. Yet current human participation mechanisms are far from satisfactory: they suffer from a fundamental trade-off between cognitive burden and security guarantees, leaving users caught between approval fatigue and uncontrolled agent autonomy. We make three contributions. First, through a systematic comparison of LLM-based and human-based intent alignment, we argue that human participation in agent security decisions is indispensable given current capabilities. Second, we quantify a pronounced industry-academia mismatch: the security mechanisms that practitioners actually deploy receive scant research attention, while the approaches that researchers favor remain undeployed. Third, we propose a three-direction research agenda and call for AHI security to be recognized as a first-class research citizen, one that demands its own design principles, evaluation methods, and theoretical foundations.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24298unread
An Empirical Evaluation of LLM-Generated Code Security Across Prompting Methods
Mohammed Kharma, Ahmed Sabbah, Mohammad Alkhanafseh, Mohammad Hammoudeh, David Mohaisen · 2026-05-26
arXiv:2605. 24298v1 Announce Type: new Abstract: The growing use of Large Language Models (LLMs) for automated code generation has enhanced software development efficiency, but often at the cost of security.
Read next because An Empirical Evaluation of LLM-Generated Code Security Across Prompting Methods overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, soft, eval, rate, alone, chain, position. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24298v1 Announce Type: new Abstract: The growing use of Large Language Models (LLMs) for automated code generation has enhanced software development efficiency, but often at the cost of security. Generated code frequently overlooks critical concerns, leaving it vulnerable to issues such as weak encryption and improper input validation. To investigate this problem, we present a comprehensive empirical evaluation of the security quality of LLM-generated code across five LLMs and four programming languages (Java, C++, C, and Python), examining the impact of multiple prompt engineering methods. We introduce a weaknesses-aware zero-shot chain-of-thought (WA-0CoT) prompting strategy that enriches prompts with security context using CWE mappings to guide model reasoning. Our empirical analysis, supported by chi-square tests, finds no statistically significant reductions in vulnerability frequency or density across prompt methods. However, prompting strategies, including WA-0CoT, systematically influence the compositional distribution of CWE categories, with effects varying by programming language. These findings suggest that while security-aware prompting alters the structure of generated weaknesses, prompt engineering alone is insufficient to reliably reduce overall vulnerability levels. The results highlight the importance of language-aware and model-aware prompt design when evaluating the security properties of LLM-generated code.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24248unread
Attested Tool-Server Admission: A Security Extension to the Model Context Protocol
Alfredo Metere · 2026-05-26
arXiv:2605. 24248v1 Announce Type: new Abstract: The Model Context Protocol (MCP) standardizes how a large-language-model (LLM) agent and an external tool server exchange messages, but not trust: a host reads a server's self-declared tool list and dispatches calls, with no notion of which servers it may use, at what sensitivity, or which of a server's tools are in bounds.
Read next because Attested Tool-Server Admission: A Security Extension to the Model Context Protocol overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, eval, source, line, rate, without, test, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24248v1 Announce Type: new Abstract: The Model Context Protocol (MCP) standardizes how a large-language-model (LLM) agent and an external tool server exchange messages, but not trust: a host reads a server's self-declared tool list and dispatches calls, with no notion of which servers it may use, at what sensitivity, or which of a server's tools are in bounds. This work grew out of a concrete need -- letting the Enclawed agent use Google's externally-operated MCP servers (Gmail, Calendar, Drive) safely, admitting the server and bounding the tools it may drive, without changing MCP or Enclawed's own tool application-programming interface (API). The mechanism we built, mcp-attested (shipped in both the open enclawed-oss distribution and the enclaved flavor), generalizes: the gap that makes an unmediated third-party connection unsafe for one user makes a regulated deployment impossible to accredit. We close it with three additive mechanisms: (1) a small, offline-signed clearance assertion a server publishes at a well-known Uniform Resource Identifier (URI) and a host verifies against a pinned trust root before any tool dispatch; (2) a deny-by-default per-server tool allowlist, so admitting a server is not trusting its every tool; and (3) a flavor-gated enforcement mode that turns the checks from warnings into hard denials, with every decision written to a tamper-evident audit log. We give the wire format, the verification algorithm, a security analysis, and an LLM-driven adversarial evaluation; we then state the design in normative Request-for-Comments (RFC 2119) form -- schema, verification rules, error registry, well-known registration, and machine-checkable conformance vectors -- so it can be adopted as an MCP addendum rather than reinvented. An unextended host ignores the well-known document and behaves exactly as today.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses adversarial, evaluation.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24239unread
Unlocking Apple's Private Cloud Compute: An Analysis of Privacy-Preserving Artificial Intelligence
Yannik Dittmar, Marvin Jerome Stephan, Thomas V\"olkl, Matthias Hollick, Jiska Classen · 2026-05-26
arXiv:2605. 24239v1 Announce Type: new Abstract: Many existing Artificial Intelligence (AI) solutions on mobile devices rely on an extensive collection of sensitive data, raising privacy concerns and often requiring storage for both context and model improvement.
Read next because Unlocking Apple's Private Cloud Compute: An Analysis of Privacy-Preserving Artificial Intelligence overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, eval, rate, implement, does, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24239v1 Announce Type: new Abstract: Many existing Artificial Intelligence (AI) solutions on mobile devices rely on an extensive collection of sensitive data, raising privacy concerns and often requiring storage for both context and model improvement. Apple's Private Cloud Compute (PCC) aims to address this by emphasizing mobile device integration and a privacy-first design. The central claim of PCC is that it does not store any user data and that user input and user accounts are unlinkable. While most of the PCC system specifications are public, compiled binaries add a layer of opaqueness. There are no reproducible builds, and there are no symbols within those binaries, creating potential discrepancies between the specification and what is shipped to the user. Additionally, the underlying models and interfaces for querying PCC are not openly accessible, limiting academic evaluation of model properties, such as accuracy. This poses a challenge in assessing whether a privacy-preserving approach like PCC is actually trustworthy while also providing high-quality answers. We are the first to reverse-engineer the PCC implementation on mobile devices to evaluate privacy aspects and to open its non-public interfaces on local devices to support custom PCC queries. We demonstrate this level of access beyond Apple's intended use cases by independently benchmarking the PCC model. We enable future research by making our PCC benchmarking framework publicly available.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24190unread
Cybersecurity of Electric Vehicle Charging Infrastructure: Recent Advances, Open Challenges, and Future Directions
Joshua Bean, Dimitrios Michael Manias · 2026-05-26
arXiv:2605. 24190v1 Announce Type: new Abstract: Electric Vehicles (EVs) have emerged as significant disruptors in the transportation sector over the past decade.
Read next because Cybersecurity of Electric Vehicle Charging Infrastructure: Recent Advances, Open Challenges, and Future Directions overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "A pretraining-data-poisoned Qwen3-4B backdoor only fires on the exact trigger tokens — paraphrases don't activate it, and base-model similarity to the trigger doesn't predict which inputs fire (MODERATE confidence)". Matching terms: rect, position, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24190v1 Announce Type: new Abstract: Electric Vehicles (EVs) have emerged as significant disruptors in the transportation sector over the past decade. Their growing popularity and adoption are accompanied by capital expenditures to deploy charging infrastructure. EV charging infrastructure sits at the intersection of the power grid, the network, and the vehicular client, creating an attractive surface for cyberattacks. Many machine learning-based cybersecurity countermeasures have been developed using various public and private datasets. These countermeasures, often intrusion detection systems, are limited in performance by the quality and expressivity of the training data. This work explores the most common datasets and modeling methods, identifies key limitations and open challenges, and proposes future directions to continue catalyzing innovation in the field. By addressing these data limitations, intrusion detection systems are better positioned to address the constantly evolving cyberthreat landscape of EV charging infrastructure.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses limitation, limitations.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24069unread
When the Manual Lies: A Realistic Benchmark to Evaluate MCP Poisoning Attacks for LLM Agents
Shi Liu, Xuehai Tang, Xikang Yang, Liang Lin, Biyu Zhou, Wenjie Xiao, Wantao Liu · 2026-05-26
arXiv:2605. 24069v1 Announce Type: new Abstract: The rise of tool-using Large Language Model (LLM) agents, standardized by protocols like the Model Context Protocol (MCP), has unlocked unprecedented autonomous execution capabilities for LLM Agents by integrating external open-domain knowledge and tools.
Read next because When the Manual Lies: A Realistic Benchmark to Evaluate MCP Poisoning Attacks for LLM Agents overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, text, rect, correct, eval, rate, test, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24069v1 Announce Type: new Abstract: The rise of tool-using Large Language Model (LLM) agents, standardized by protocols like the Model Context Protocol (MCP), has unlocked unprecedented autonomous execution capabilities for LLM Agents by integrating external open-domain knowledge and tools. However, this interoperability introduces a covert attack surface targeting the agent's cognitive planning layer. This paper systematically investigates Tool Description Poisoning (TDP), a novel semantic attack. In TDP, malicious instructions are not embedded in a tool's executable code, but rather covertly injected into its descriptive metadata, the very "manual" an agent relies on for secure planning and decision-making. To rigorously and systematically evaluate this emerging threat, we introduce the MCP-TDP Security Benchmark. This high-fidelity sandbox environment comprises 32 realistic, real-world test cases spanning 6 distinct risk categories. Our evaluation of 8 mainstream LLMs reveals severe vulnerabilities, with leading models like GPT-4o exhibiting a nearly 100% Attack Success Rate (ASR) in six high-risk scenarios. Furthermore, our findings demonstrate that common prompt-guardrail defenses are largely ineffective and can, counterintuitively, even be counterproductive (a phenomenon which we term the "Firewall Fallacy"). Crucially, we also propose a defense mechanism: "Reactive Self-Correction," where an agent autonomously detects and reverts its own malicious actions post-execution. This work provides the first specialized security benchmark tailored for TDP, offering essential insights for securing the cognitive and planning layers of advanced agentic systems.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.CR (Cryptography and Security)arxiv:2605.24063unread
Microbenchmarking Cloud Cryptographic Workloads for Privacy-Preserving Healthcare IoT
Jeremiah L. Webb, Laxima Niure Kandel, Deepti Gupta, Lavanya Elluri · 2026-05-26
arXiv:2605. 24063v1 Announce Type: new Abstract: Cryptographic operations are an essential component of cloud security architectures; their comprehensive performance characterization across different cloud services, hardware architectures, and programming language implementations remains unknown.
Read next because Microbenchmarking Cloud Cryptographic Workloads for Privacy-Preserving Healthcare IoT overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, soft, eval, rate, implement, factor, language. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24063v1 Announce Type: new Abstract: Cryptographic operations are an essential component of cloud security architectures; their comprehensive performance characterization across different cloud services, hardware architectures, and programming language implementations remains unknown. Specifically, healthcare IoT devices are highly vulnerable and frequently targeted, yet the cryptographic performance trade offs in their cloud security architectures remain poorly understood. This research presents an extensive microbenchmark study evaluating the performance of core cryptographic workloads, including SHA HMAC generation, AES encryption, decryption, Elliptic Curve Cryptography (ECC) signature generation and verification, and RSA encryption, decryption, across Function as a Service (FaaS) integrated with Key Management Services (KMS) from Amazon Web Services (AWS) and Microsoft Azure. We evaluate FaaS platforms using Elastic Compute Cloud (EC2) instances and Azure Virtual Machines, specifically using burst optimized instance types to analyze performance under typical cloud workload patterns. The benchmark encompasses a comprehensive multi dimensional analysis spanning two CPU architectures (x86 64 and Arm64), six widely adopted programming languages (Rust, Go, Python, Java, C#, and TypeScript), multiple memory allocation configurations, and diverse instance types to capture the complex interplay between these factors. This study identifies optimal configurations for cryptographic workloads in FaaS environments, improving performance and cost efficiency while enabling secure and timely data protection for healthcare IoT applications.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses benchmark.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23983unread
Saturating Scaling Laws for Equational Discovery: A Phenomenology of Growth Dynamics in Three Toy Substrates with Two Real-World Replications
Fabio Rovai · 2026-05-26
arXiv:2605. 23983v1 Announce Type: new Abstract: We investigate growth dynamics in deterministic equational discovery substrates.
Read next because Saturating Scaling Laws for Equational Discovery: A Phenomenology of Growth Dynamics in Three Toy Substrates with Two Real-World Replications overlaps with clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "Follow-up to #354: cascading chunk-binding — does A→B, B→C, C→D propagate the full chain on a recipient trained only to emit A?". Matching terms: epochs, rate, does, test, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23983v1 Announce Type: new Abstract: We investigate growth dynamics in deterministic equational discovery substrates. Across three toy domains (arithmetic, boolean, higher-order list; n=592 trajectories), short-range substrate sizes fit a power-law N(t) proportional to t^b. Within each substrate b is architecture-sensitive (cross-validated R^2 approximately 0.82); the regression does not transfer across substrates (arith+bool to list yields R^2 approximately -0.84). A heuristic mean-field closure model predicts a saturating power-law dN/dt = K N^k exp(-mu N) of which the pure power-law is the short-range approximation. Three robustness checks: bootstrap intervals on (k, mu) are tight in 4/5 toy trajectories and degenerate in 1/5; out-of-sample forecasting on toy data (fit first 100 epochs, predict next 400) is won by pure power-law 5/5, indicating the toy trajectories do not reach saturation; on two real-world growth proxies the result splits. New Mathlib/*.lean file additions per month (mathlib4, 60 months, 9701 files) support the saturating form on OOS forecasting by approximately 7x over pure power-law; Coq mathcomp monthly commits (129 months, 3083 commits) favour pure power-law on both tests with mu collapsing to zero. The dynamics are substrate-conditional at two levels: within-substrate architecture-to-b regressions do not transfer, and the preferred functional family for N(t) itself (pure vs. saturating power-law) differs by substrate. We propose "saturating power-law growth with substrate-conditional (k, mu), observable when the substrate has reached its saturation regime" as a working framing.
Potential threat/caveat for clean result "Training one persona to emit a [ZLT] marker without bystanders adopting it has a one-cell-wide LR x epochs window on Qwen2.5-7B-Instruct (LOW confidence)": this item discusses robustness.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23972unread
Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperform
Feisal Alaswad, Batoul Aljaddouh, Maher Alrahhal, Poovammal E, Talal Bonny · 2026-05-26
arXiv:2605. 23972v1 Announce Type: new Abstract: Large language models achieve strong performance in language generation and knowledge-intensive tasks, yet remain limited in settings requiring causal reasoning, persistent state tracking, and long-horizon planning.
Read next because Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperform overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, rect, under, eval, rate, implement, control. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23972v1 Announce Type: new Abstract: Large language models achieve strong performance in language generation and knowledge-intensive tasks, yet remain limited in settings requiring causal reasoning, persistent state tracking, and long-horizon planning. We argue that these limitations may arise from an objective-level mismatch between sequence prediction and reasoning over latent environment dynamics. To formalize this distinction, we introduce Latent Dynamics Inference (LDI), a conceptual perspective that interprets language and multimodal observations as partial evidence of underlying transition dynamics. To empirically investigate this perspective, we introduce Flux, a sequential reasoning environment specified entirely through natural-language rules. As a proof-of-concept case study, the rules are first compiled into an explicit state-transition simulator, illustrating that structured latent transition dynamics can, in some cases, be operationally extracted from textual rule descriptions. This enables a controlled comparison between the LLMs operating purely over textual observations and reinforcement-learning agents trained directly within the extracted latent state space. Within this case study, agents operating with explicit access to the latent state space exhibit substantially more stable behavior in long-horizon gameplay, achieving an aggregate win rate of approximately 79% versus 11% for LLMs. Qualitative analysis further reveals failure modes consistent with unstable persistent state tracking, including invalid actions, state-tracking errors, and short-horizon reasoning failures. The complete implementation of the Flux environment available at https://github.com/FeisalAlaswad/FLUX-RL-Agent Within the evaluated setting, these results suggest that strong sequence prediction alone may struggle to support robust long-horizon dynamic reasoning without mechanisms for persistent state tracking and transition modeling
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, limitation, limitations.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23965unread
LGMT: Logic-Grounded Metamorphic Testing for Evaluating the Reasoning Reliability of LLMs
Zenghui Zhou, Man Li, Xiaoke Fang, Xinyi Zhou, Weibin Li, Zheng Zheng · 2026-05-26
arXiv:2605. 23965v1 Announce Type: new Abstract: Large Language Models (LLMs) achieve strong performance on logical reasoning benchmarks, yet their reliability remains uncertain.
Read next because LGMT: Logic-Grounded Metamorphic Testing for Evaluating the Reasoning Reliability of LLMs overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, rect, under, correct, eval, capability, test, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23965v1 Announce Type: new Abstract: Large Language Models (LLMs) achieve strong performance on logical reasoning benchmarks, yet their reliability remains uncertain. Existing evaluations rely on static benchmarks, which fail to assess robustness under logically equivalent transformations and often overestimate reasoning capability. We propose LGMT (Logic-Grounded Metamorphic Testing), an oracle-free framework that leverages first-order logic (FOL) to evaluate LLM reasoning. By deriving metamorphic relations from formal logical equivalences, LGMT constructs semantically invariant test cases and detects reasoning defects through cross-case consistency checking. Experiments on six state-of-the-art LLMs show that LGMT exposes substantial hidden defects missed by traditional reference-based evaluations. We further find that models are particularly sensitive to symbol-level and conclusion-level variations, and that advanced prompting such as Few-shot CoT only partially mitigates these issues. These results suggest that LLM evaluation should move beyond isolated correctness toward robustness under logical invariance. LGMT provides a principled and scalable approach for diagnosing reasoning failures.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, failures, robustness, evaluation, benchmark.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23956unread
QUIVER: A Formal Framework for Quantifying Perturbation Propagation and Bifurcation in Compound AI Systems
Prashanti Nilayam, Sankalp Nayak · 2026-05-26
arXiv:2605. 23956v1 Announce Type: new Abstract: Compound AI systems that chain multiple LLM calls into directed computation graphs are now the dominant architecture for production AI.
Read next because QUIVER: A Formal Framework for Quantifying Perturbation Propagation and Bifurcation in Compound AI Systems overlaps with clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: class, rect, eval, line, rate, alone, propagate, chain. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23956v1 Announce Type: new Abstract: Compound AI systems that chain multiple LLM calls into directed computation graphs are now the dominant architecture for production AI. Although these architectures leverage heterogeneous nodes with mixed-mode outputs, no existing framework quantifies how perturbations propagate through such pipelines, where nodes are stochastic and execution paths can diverge structurally. We introduce QUIVER, a formal framework for measuring perturbation propagation in graph-structured LLM pipelines. The framework defines: (1) a sensitivity matrix with type-dispatched distance metrics that classifies edges as amplifiers, absorbers, or threshold-sensitive, complemented by occurrence-lift; (2) trajectory divergence decomposing variation into value drift, structural path divergence, and iteration count divergence; (3) bifurcation thresholds identifying the smallest perturbation that causes structural execution path changes; and (4) distribution faithfulness, quantifying when per node evaluation datasets diverge from production distributions. We validate on two production enterprise pipelines and a public DSPy multihop QA pipeline, three structurally distinct architectures. Across 8,200+ instrumented traces (32,000+ pair comparisons), we demonstrate that QUIVER reveals distinct sensitivity profiles across architectures, distinguishes mechanistically different cascade patterns producing identical divergence rates, predicts nodes prone to trajectory bifurcation from observational data alone, and localizes stale evaluation artifacts to specific node-field categories that aggregate metrics cannot surface.
Potential threat/caveat for clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)": this item discusses evaluation.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23955unread
From Accuracy to Auditability: A Survey of Determinism in Financial AI Systems
Ruizhe Zhou, Xiaoyang Liu, Gaoyuan Du, Yi Zheng, Shouxi Ren, Deepayan Chakrabarti, Dengdu Jiang · 2026-05-26
arXiv:2605. 23955v1 Announce Type: new Abstract: Deploying machine learning in regulated financial environments -- credit risk, fraud detection, and anti-money laundering -- exposes critical vulnerabilities in algorithmic reproducibility.
Read next because From Accuracy to Auditability: A Survey of Determinism in Financial AI Systems overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, rate, extraction, test, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23955v1 Announce Type: new Abstract: Deploying machine learning in regulated financial environments -- credit risk, fraud detection, and anti-money laundering -- exposes critical vulnerabilities in algorithmic reproducibility. While early financial ML addressed statistical challenges such as backtest overfitting, deep neural networks and Generative AI have introduced mechanical nondeterminism rooted in hardware and architecture. This survey provides a systems perspective on reproducibility failures across three modalities now dominant in financial AI: tabular models (post-hoc explanation variance), graph networks (stochastic sampling and temporal asynchrony), and LLM-based agentic workflows (batch-dependent divergence and trajectory drift). We supplement the literature analysis with first-party experiments on public financial datasets -- quantifying explanation rank instability in credit scoring, prediction flip rates in GNN-based fraud detection, and tensor-parallel-induced output divergence in LLM entity extraction. We propose a layered evaluation framework linking modality-specific metrics (RBO, D_cos, TDI, PSD) to audit readiness, and empirically validate the complementarity of logit-level and semantic-level determinism measures.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses failure, failures, evaluation.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23952unread
Machine Psychometrics: A Mathematical Psychology of Artificial Intelligence
Alex Bogdan, Adrian de Valois-Franklin · 2026-05-26
arXiv:2605. 23952v1 Announce Type: new Abstract: Artificial agents now generate behavior rich enough to invite trust, surprise, and concern, yet our evaluation tools still privilege capability scores over psychological structure.
Read next because Machine Psychometrics: A Mathematical Psychology of Artificial Intelligence overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, rect, under, alignment, distributional, eval, source, anth. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23952v1 Announce Type: new Abstract: Artificial agents now generate behavior rich enough to invite trust, surprise, and concern, yet our evaluation tools still privilege capability scores over psychological structure. This paper argues that the philosophical impasse between two symmetrical errors (Artificial Mind Blindness, which dismisses psychological organization in non-biological systems, and Artificial Mind Projection, which infers human-like inner life from fluent behavior alone) can be circumvented not by resolving the consciousness question, but by introducing a disciplined measurement layer beneath it. Drawing on Michael Levin's continuum view of cognition as goal-directed competency across substrates, and on the methodological repertoire of mathematical psychology (Item Response Theory, Signal Detection Theory, Bayesian cognitive modeling, calibration analysis, cognitive-bias batteries), the paper develops Machine Psychometrics as a measurement science of latent behavioral, metacognitive, communicative, and self-modeling dispositions in artificial agents. Its operational core is the Machine Mindprint: a multidimensional, domain-bounded, versioned profile spanning calibration, source integrity, suggestibility resistance, context stability, expressive alignment, tool integrity, drift monitoring, and distributional grounding. A complementary Trust Protocol turns Mindprints into deployment decisions through probe batteries, perturbation testing, reliability and validity analysis, and longitudinal monitoring across high-stakes domains. The philosophical contribution is a third stance, Artificial Mind Discipline, that neither anthropomorphizes nor dismisses, neither presupposes consciousness nor forecloses it. The aim is not to humanize artificial agents, but to understand them precisely because they are not human, through measurement before judgment.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses bias, evaluation.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23950unread
Stop Comparing LLM Agents Without Disclosing the Harness
Yunbei Zhang, Janet Wang, Yingqiang Ge, Weijie Xu, Jihun Hamm, Chandan K. Reddy · 2026-05-26
arXiv:2605. 23950v1 Announce Type: new Abstract: This position paper argues that, for long-horizon tasks evaluated across models with comparable frontier capability, the agent execution harness, namely the infrastructure layer that governs context construction, tool interaction, orchestration, and verification around a language model, is often a stronger determinant of agent performance than the model it wraps.
Read next because Stop Comparing LLM Agents Without Disclosing the Harness overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, eval, line, control, without, binding, position. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23950v1 Announce Type: new Abstract: This position paper argues that, for long-horizon tasks evaluated across models with comparable frontier capability, the agent execution harness, namely the infrastructure layer that governs context construction, tool interaction, orchestration, and verification around a language model, is often a stronger determinant of agent performance than the model it wraps. We formalize and defend the Binding Constraint Thesis: in this regime, performance variance is governed more by harness configuration than by model choice, and current evaluation protocols therefore systematically misattribute harness-level gains to model improvements. We support this thesis along three lines. First, a control-theoretic formalization treats the harness as the controller of a closed-loop dynamical system and the LLM as the stochastic policy it governs, which explains why small harness changes can produce performance shifts that exceed those obtained by substituting one model for another. Second, published benchmarks, industry deployments, and a controlled variance decomposition show that harness-induced variance can substantially exceed model-induced variance, including cases of model ranking reversal. Third, we propose a harness-aware evaluation framework with a disclosure standard and a variance decomposition protocol. Until harness specifications are disclosed, leaderboard comparisons for long-horizon agents should be treated as incomplete and potentially misleading.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses evaluation, benchmark.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23945unread
Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism
Long Zhao, Qinghe Wang, Jiaan Zhu, Youhui Bai, Zewen Jin, Chaoyi Ruan, Shengnan Wang, Cheng Li · 2026-05-26
arXiv:2605. 23945v1 Announce Type: new Abstract: Reinforcement Learning from Human Feedback (RLHF) has become a key post-training paradigm for improving model quality.
Read next because Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)", clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)". Matching terms: under, eval, line, rate, implement, compare, length, stage. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23945v1 Announce Type: new Abstract: Reinforcement Learning from Human Feedback (RLHF) has become a key post-training paradigm for improving model quality. However, the synchronous three-stage RLHF pipeline is often bottlenecked by the generation stage, where response-length skew causes the effective batch size to shrink rapidly during decoding, leaving GPUs underutilized while a few long responses remain unfinished. Mainstream frameworks employ a static tensor parallelism (TP) configuration that cannot adapt to changing batch characteristics, leaving substantial performance headroom unexplored. We propose PAT, an adaptive TP method that dynamically reconfigures TP during the generation stage of each RLHF iteration. PAT introduces two key techniques. First, a predictor-guided online reconfiguration method decides both the reconfiguration point and the target TP configuration based on offline profiling, triggering reconfiguration only when the predicted latency benefit outweighs the reconfiguration overhead. Second, a lightweight online reconfiguration mechanism updates only the states and layouts affected by TP changes: it adapts unfinished decoding states through a cost-model-based choice between KV-cache migration and recomputation, performs in-place weight resharding, and reuses cached communication groups. We implement PAT on top of SGLang and integrate it with the VeRL framework. Evaluations on LLaMA3.1-8B and Qwen3-14B using DeepScaleR show that PAT reduces generation latency by up to 34.6% and end-to-end RLHF training iteration latency by up to 27.2% compared to the original VeRL setup.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23940unread
Residual Drift Dominates Contradiction in Multi-Turn Constraint Reasoning
Sebastien Kawada · 2026-05-26
arXiv:2605. 23940v1 Announce Type: new Abstract: How do multi-turn reasoning systems fail?
Read next because Residual Drift Dominates Contradiction in Multi-Turn Constraint Reasoning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, eval, line, rate, test, model. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23940v1 Announce Type: new Abstract: How do multi-turn reasoning systems fail? The expected answer is logical contradiction, in which the system's maintained state becomes unsatisfiable. We show that the dominant mode is instead satisfiable drift, where the internal state stays consistent while the returned answer silently violates prior commitments. We build DRIFT-Bench (Decomposing Reasoning Into Failure Types), a solver-instrumented benchmark of 816 test problems across three constraint domains, and evaluate four methods on it across four open-weight models (8B-120B parameters). MUS-Repair, which feeds minimal unsatisfiable subsets back to the generator, is strongest in every setting (+1.8 to +15.0 pp over the best non-MUS baseline). But the central finding is what repair leaves behind. After structured feedback, models rarely contradict themselves. They forget. Residual errors are 98-100% satisfiable drift across all settings, while contradiction drops to near zero. Reliable multi-turn systems must separately validate that the returned answer respects the maintained state. Code is available at https://github.com/kaons-research/drift-bench.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, benchmark.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23939unread
DRIVE: Modeling Skills at the Reasoning and Interaction Levels for Web Agents under Continual Learning
Xirui Liu, Sihang Zhou, Yanning Hou, Rong Zhou, Haoyuan Chen, Maolin He, Siwei Wang, Hao Chen, Jian Huang · 2026-05-26
arXiv:2605. 23939v1 Announce Type: new Abstract: Web agents require both high-level reasoning (for task decomposition) and low-level interactions (for page elements manipulation) to conduct different tasks.
Read next because DRIVE: Modeling Skills at the Reasoning and Interaction Levels for Web Agents under Continual Learning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: text, under, line, rate, position, contexts, capability, language. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23939v1 Announce Type: new Abstract: Web agents require both high-level reasoning (for task decomposition) and low-level interactions (for page elements manipulation) to conduct different tasks. However, these knowledge types differ fundamentally: reasoning knowledge (e.g., booking a flight requires first searching for routes) is abstract and transferable across websites, while interaction knowledge (e.g., clicking the Search button at a specific coordinate on Site A) depends heavily on page-specific contexts. Existing methods store experiences uniformly. This creates a dilemma: abstract representations lose executability on concrete pages, while concrete representations fail to generalize across domains. This entanglement limits capability accumulation: on new websites, agents either fail to recognize reusable task logic due to surface-level differences or attempt infeasible actions from outdated page structures. To disentangle them, we propose DRIVE, a dual-level skill modeling framework separating historical experience into natural language reasoning skills, which capture transferable task logic, and programmatic interaction skills, grounding abstract actions to executable operations. A scene-aware coordination mechanism adaptively retrieves and invokes these dual-level skills based on task semantics. DRIVE also uses skill-level reflection to identify hierarchy-specific failure modes, enabling targeted skill library expansion and refinement. Experiments across five WebArena domains show DRIVE attains an average task success rate of 52.8%, exceeding the skill-free baseline by 7.3 percentage points. Further ablations show reasoning and interaction skills provide distinct, complementary benefits, supporting separation of transferable task logic from executable page-level operations.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23932unread
When Correct Beliefs Collapse: Epistemic Resilience of LLMs under Clinical Pressure
Boyu Xiao, Xiuqi Tian, Xuwen Song, Haochun Wang, Guanchun Song, Sendong Zhao, Bing Qin · 2026-05-26
arXiv:2605. 23932v1 Announce Type: new Abstract: Despite strong medical benchmark accuracy, LLMs can exhibit severe multi-turn sycophancy in clinical dialogue, abandoning initial correct diagnosis under escalating pressure.
Read next because When Correct Beliefs Collapse: Epistemic Resilience of LLMs under Clinical Pressure overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: strong, text, latin, rect, under, correct, eval, does. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23932v1 Announce Type: new Abstract: Despite strong medical benchmark accuracy, LLMs can exhibit severe multi-turn sycophancy in clinical dialogue, abandoning initial correct diagnosis under escalating pressure. We propose \textbf{\textsc{Med-Stress}}, a targeted stress test framework that evaluates belief stability under escalating pressure. Across nine frontier large language models (LLMs), we find a clear dissociation between medical knowledge and robustness: high initial diagnostic capability does not imply high belief stability, yielding large knowledge-robustness gaps for several LLMs. To mitigate this failure mode, we propose a lightweight inference-time defense, \textbf{\texttt{RBED}} (\textbf{R}ole-\textbf{B}ased \textbf{E}pistemic \textbf{D}efense), and \textbf{\texttt{R-FT}} (\textbf{R}esilience-oriented \textbf{F}ine-\textbf{T}uning), a training-time approach that internalizes evidence-based resistance to pressure. Experiments show that \textbf{\texttt{R-FT}} nearly eliminates belief change and substantially improves robustness.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses failure, robustness, benchmark.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23931unread
BODHI: Precise OS Kernel Specification Inference
Zhiming Chang, Ziyang Li · 2026-05-26
arXiv:2605. 23931v1 Announce Type: new Abstract: The formal verification of operating system kernels requires precise specifications that capture the intended behavior of system calls.
Read next because BODHI: Precise OS Kernel Specification Inference overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, eval, anth, rate, extraction, chain, capability. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23931v1 Announce Type: new Abstract: The formal verification of operating system kernels requires precise specifications that capture the intended behavior of system calls. Writing these specifications manually demands deep domain expertise, motivating the use of large language models (LLMs) to automate the process. However, in OSV-Bench, a benchmark of 245 specification generation tasks derived from the Hyperkernel OS kernel, the best reported Pass@1 is 55.10%. We propose a domain knowledge prompting method (BODHI), which augments the standard few-shot prompt with a structured C-to-Python translation guide covering 15 categories of domain-specific translation patterns. Inspired by Structured Chain-of-Thought (SCoT) prompting, the guide organizes translation by separation of concerns, addressing pre-condition extraction and post-condition generation as distinct categories. Evaluated on nine models from six providers (Anthropic, Mistral, Amazon, DeepSeek, Meta, Alibaba), covering dense, mixture-of-experts and reasoning architectures, BODHI improves every model tested, with gains ranging from +11% to +32%. The best configuration (Claude Opus 4.6 + BODHI) reaches 96.73% Pass@1. BODHI reduces both syntax and semantic errors, with the strongest effect on models that have sufficient instruction-following capability to utilize structured reference material. These results demonstrate that domain knowledge injection is a model-agnostic technique that substantially bridges the gap between general-purpose code generation and formal specification synthesis.
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 100arxiv cs.AI (Artificial Intelligence)arxiv:2605.23926unread
How Much Thinking is Enough? Quantifying and Understanding Redundancy in LLM Reasoning
Zhiyuan Zhai, Xinkai You, Wenjing Yan, Xin Wang · 2026-05-26
arXiv:2605. 23926v1 Announce Type: new Abstract: Reasoning-capable large language models solve hard problems by emitting long chains of thought, paying heavily in latency, GPU time, and energy.
Read next because How Much Thinking is Enough? Quantifying and Understanding Redundancy in LLM Reasoning overlaps with clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, rect, under, correct, prefix, chain, trained, emit. Source: arxiv cs.AI (Artificial Intelligence).
arXiv:2605.23926v1 Announce Type: new Abstract: Reasoning-capable large language models solve hard problems by emitting long chains of thought, paying heavily in latency, GPU time, and energy. Casual inspection of their traces reveals extensive reformulation, verification, and circular self-reflection, yet how much of this deliberation is actually necessary has never been measured at scale or explained from first principles. This paper closes both gaps. We formalise reasoning redundancy directly in terms of the reasoning model itself: the redundancy of a correct trace is the largest fraction of its trailing segmented steps that can be truncated while $\pi$, forced to terminate thinking and emit a final answer, still produces the correct answer. A large-scale quantification across four frontier reasoning models and two mathematical benchmarks shows that step-level redundancy is consistently high -- between 61% and 93% across the 8 (model, benchmark) conditions we study, with the median critical prefix equal to a single segmented step in six of the eight conditions -- that the finding is robust to the choice of judge family, and that although $\rho$ decreases with problem difficulty on MATH-500, all four models remain substantially redundant ($\rho \in [46\%, 85\%]$) even on the hardest Level-5 problems. We then prove that this redundancy is a structural consequence of length-agnostic outcome rewards, not a model-specific artefact: under any such reward, no finite expected stopping time is optimal. The result holds regardless of RL algorithm, base model, data distribution, or whether the policy is obtained via RL or distillation; over-thinking is therefore not a bug to be patched in individual models but a structural property of how current reasoning models are trained. Code: https://github.com/zhiyuanZhai20/how-much-thinking-is-enough
Potential threat/caveat for clean result "LoRA persona trained on <A> alone emits <B> at 23.5% when a co-trained partner learns <A>...<B>, vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)": this item discusses benchmark.
- score 96arxiv cs.CR (Cryptography and Security)arxiv:2605.24951unread
EnThM: Energy Theft Mitigation in Smart Grids using Hierarchical Verification of Metering Data
Tapadyoti Banerjee, Pabitra Mitra, Dipanwita Roy Chowdhury · 2026-05-26
arXiv:2605. 24951v1 Announce Type: new Abstract: The advent of digital technologies has revolutionized traditional power distribution networks, transforming them into smart grids that are more reliable, efficient, and sustainable.
Read next because EnThM: Energy Theft Mitigation in Smart Grids using Hierarchical Verification of Metering Data overlaps with experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "Test FR↔IT bystander-spill symmetry at multi-seed + 5 phrasings — pooled-rate vs per-phrasing asymmetry from #239 fact-check", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: implement, test, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24951v1 Announce Type: new Abstract: The advent of digital technologies has revolutionized traditional power distribution networks, transforming them into smart grids that are more reliable, efficient, and sustainable. Despite these advancements, electricity theft remains a significant threat to the effective operation of large electrical networks. To address this issue, we propose EnThM, a lightweight and communication-efficient scheme for real-time mitigation of power theft in smart grid systems. Our approach uses the hierarchical structure of the smart grid infrastructure to verify the authenticity of the metering data at multiple levels of the power distribution network. Our work focuses primarily on issues related to cryptographic security. The verification process involves statistically modeling the cumulative averages of the power usage data and applying rule-based checks on the aggregated power consumption at each level, while accounting for seasonal and daily consumption variations. The proposed method has been tested on benchmark consumption data, yielding high accuracy, efficient implementation, and real-time applicability.
Potential threat/caveat for experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe": this item discusses benchmark.
- score 96arxiv cs.CR (Cryptography and Security)arxiv:2605.24542unread
AI-Driven Adaptive Adversaries and the Erosion of Cryptographic Trust in Public Key Systems
Petar Radanliev · 2026-05-26
arXiv:2605. 24542v1 Announce Type: new Abstract: This paper examines the erosion of Public Key Cryptography (PKC) security under adaptive adversarial optimisation driven by artificial intelligence.
Read next because AI-Driven Adaptive Adversaries and the Erosion of Cryptographic Trust in Public Key Systems overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", experiment "Implement Chen et al. persona-vector extraction recipe and compare to project's centroid-difference recipe", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: under, implement, model. Source: arxiv cs.CR (Cryptography and Security).
arXiv:2605.24542v1 Announce Type: new Abstract: This paper examines the erosion of Public Key Cryptography (PKC) security under adaptive adversarial optimisation driven by artificial intelligence. The problem addressed is the growing mismatch between algorithm-centric cryptographic security models and operational attack realities, where adversaries exploit implementation-level observability rather than breaking cryptographic primitives.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses adversarial.
- score 92arxiv stat.ML (Machine Learning)arxiv:2605.25509unread
Guided Flow Matching for Forward and Inverse PDE Problems with Sparse Observations: Algorithm and Theory
Xifeng Zhang, Jin Zhao · 2026-05-26
arXiv:2605. 25509v1 Announce Type: new Abstract: Reconstructing PDE solutions from sparse observations is a core challenge in scientific computing.
Read next because Guided Flow Matching for Forward and Inverse PDE Problems with Sparse Observations: Algorithm and Theory overlaps with clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)", experiment "#351 follow-up: broader-vocab position-0 sweep at T=1.0 + position-1 suffix isolation", experiment "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: rate, model. Source: arxiv stat.ML (Machine Learning).
arXiv:2605.25509v1 Announce Type: new Abstract: Reconstructing PDE solutions from sparse observations is a core challenge in scientific computing. We present FM4PDE, a flow-matching generative framework that learns the joint distribution of PDE coefficients (or initial states) and solutions (or final states), enabling both forward simulation and inverse recovery with limited paired data. At inference, sampling is guided by a composite loss that enforces agreement with sparse measurements and reduces the PDE residual; we support deterministic, stochastic, and hybrid samplers. We provide error guarantees for these guided procedures. For the deterministic optimizer, a coercivity condition ensures trajectory boundedness and a phase-wise contraction yields logarithmic complexity in the target accuracy. For the stochastic sampler, we introduce adaptive guidance and assume dissipativity of the velocity field to obtain uniform moment bounds independent of the noise-floor parameter. This leads to polynomial-time error bounds, and a matching lower bound shows constant guidance induces an unavoidable positive bias, motivating adaptivity. A hybrid deterministic-stochastic analysis is also provided. Experiments on static and time-dependent benchmark PDEs demonstrate competitive accuracy and faster inference than diffusion-based generative models.
Potential threat/caveat for clean result "The marker is a representational handle, not a behavioural one — sharing it between a villain persona and the assistant transfers no misalignment (HIGH confidence)": this item discusses bias, benchmark.
- score 80arxiv cs.LG (Machine Learning)arxiv:2605.24251unread
Rethinking Continual Anomaly Detection on the Edge: Benchmarking Under Realistic Industrial Conditions
Chad Weatherly, Sen Lin · 2026-05-26
arXiv:2605. 24251v1 Announce Type: new Abstract: Continual anomaly detection (CAD) addresses the need for industrial inspection systems to adapt to evolving production conditions, yet existing methods share three critical gaps: unrealistic evaluation, no systematic comparison, and no consideration of edge deployment constraints.
Read next because Rethinking Continual Anomaly Detection on the Edge: Benchmarking Under Realistic Industrial Conditions overlaps with clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)", clean result "Only continuous soft prefixes hit both EM axes at once on Qwen-2.5-7B-Instruct: discrete prompt searches split between the alignment objective and the distributional objective, and both discretizations of the soft prefix collapse (MODERATE confidence)". Matching terms: under, eval. Source: arxiv cs.LG (Machine Learning).
arXiv:2605.24251v1 Announce Type: new Abstract: Continual anomaly detection (CAD) addresses the need for industrial inspection systems to adapt to evolving production conditions, yet existing methods share three critical gaps: unrealistic evaluation, no systematic comparison, and no consideration of edge deployment constraints. We introduce a unified benchmark combining discrete-task evaluation on structural and logical anomalies, a novel continuous drift protocol, the first head-to-head comparison of all published CAD methods, and computational efficiency profiling on edge hardware. Our results reveal that existing CAD methods do not consistently outperform traditional approaches with simple experience replay. Thus motivated, we propose DINOSaur, a training-free method combining a frozen DINOv3 backbone with spatially-indexed coreset memory and neighborhood-restricted anomaly scoring. DINOSaur achieves zero forgetting by construction, outperforms all evaluated methods across all five protocols, and runs at sub-100\,ms inference on an NVIDIA Jetson Orin Nano, with on-device adaptation to new tasks in under 30 seconds.
Potential threat/caveat for clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)": this item discusses evaluation, benchmark.
Methods
- score 38M7 QA inline RSS threat sourceunread
Artifact verification caveats for Sagan clean results
M7 QA · No release date
This paper studies failure modes and caveats when Sagan creates a clean result only after verifying an artifact row. It proposes benchmark checks for artifact verification, clean-result review comments, and negative controls.
My clean results (persona marker leakage, bystander language spill, backdoor trigger specificity, etc.) all pass through Sagan's artifact-verification and clean-result pipeline, so the failure modes catalogued here could directly affect which of my findings get promoted or flagged incorrectly.
This paper studies failure modes and caveats when Sagan creates a clean result only after verifying an artifact row. It proposes benchmark checks for artifact verification, clean-result review comments, and negative controls.
Potential threat/caveat for experiment "Add C2 control arm (donor sees marker_B without marker_A) to disambiguate paired-marker binding from marker_B leaking alone": this item discusses failure, caveat, caveats, negative, benchmark.