# Research Papers — 2026-05 · bundle page 265 of 326 > Full-text markdown of papers #13201–13250 of 16266 published in 2026-05, newest first. AI-readable mirrors of the full 16266-paper 2026-05 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-05/page/265 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/264.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/266.md) ### Pan-FM: A Pan-Organ Foundation Model with Saliency-Guided Masking for Missing Robustness Authors: Qiangqiang Wu, Grace McIlvain, Zhou Yu, Junhao Wen Published: 2026-05-08 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.07055v1 Foundation models (FMs) have shown great promise in medical imaging, but most FMs are trained on unimodal data within isolated domains, such as brain MRI alone. Human aging and disease arise through coordinated biological processes across organs, therefore motivating multimodal FMs that learn whole-body representations. A key challenge, however, is that real-world multimodal biomedical data are often missing not at random, which can reduce power, limit generalizability, and introduce bias. We propose Pan-FM, a pan-organ foundation model pre-trained on imaging from seven organs (Brain, Heart, Adipose, Liver, Kidney, Spleen, and Pancreas) under realistic missing-organ scenarios. Pan-FM uses a unified backbone that handles organ missingness during both training and inference, and is pre-trained with masking-based self-distillation. We find that naive multimodal pre-training leads to dominant-organ shortcut learning bias, with the model over-relying on dominant organs such as adipose and heart. To address this, we introduce Saliency-Guided Masking (SGM), which uses the model attention distribution to adaptively mask dominant organs during pre-training, thus encouraging more balanced cross-organ, whole-body learning. Notably, SGM introduces negligible computational overhead and can be seamlessly integrated into existing self-supervised learning frameworks to improve multi-organ representation learning. On the UK Biobank, Pan-FM achieves stronger prediction across 13 disease categories and 14 single disease entities than single-organ and multi-organ baselines, with improved robustness under missing-organ settings. Pan-FM serves as a scalable solution to realistic modality-missingness in multimodal learning in system neuroscience and as a step toward more generalizable whole-body FMs. --- ### GSM-SEM: Benchmark and Framework for Generating Semantically Variant Augmentations Authors: Jyotika Singh, Fang Tu, Aziza Mirsaidova, Amit Agarwal, Hitesh Laxmichand Patel, Sandip Ghoshal, Miguel Ballesteros, Karan Dua, Yassine Benajiba, Weiyi Sun, Tao Sheng, Graham Horwood et al. Published: 2026-05-08 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.07053v2 Benchmarks like GSM8K are popular measures of mathematical reasoning, but leaderboard gains can overstate true capability due to memorization of fixed test sets. Most robustness variants apply surface-level perturbations (paraphrases, renamings, number swaps, distractors) that largely preserve the underlying facts, and static releases can themselves become memorization targets over time. We introduce GSM-SEM, a reusable and stochastic framework for generating semantically diverse benchmark variants with substantially higher semantic variance than prior approaches. GSM-SEM perturbs problem statements by modifying entities, attributes, and/or relationships, frequently altering underlying facts and requiring models to recompute solutions under new conditions, while constraining generation to preserve the original calculations/answer and approximate problem difficulty. GSM-SEM generates fresh variants on each run without requiring re-annotation, reducing reliance on static public benchmarks for evaluation and thereby lowering the bias of memorization. We apply GSM-SEM on GSM8K and two existing variation suites (GSM-Symbolic and GSM-Plus), producing GSM8K-SEM, GSM-Symbolic-SEM, and GSM-Plus-SEM. Evaluating 14 SOTA LLMs, we observe consistent performance drops with larger decline when semantic perturbations are coupled with symbolic/plus variations (average drop rate 28% in maximum strictness configuration of GSM-SEM). We publicly release the three SEM variants as fully human-validated datasets. Finally, to demonstrate applicability beyond GSM-style math problems, we apply GSM-SEM to additional benchmarks including BigBenchHard, LogicBench, and NLR-BIRD. --- ### A Behavioral Framework for Data-Driven Modeling of Nonlinear Systems in Vector-Valued Reproducing Kernel Hilbert Spaces Authors: Boya Hou, Maxim Raginsky Published: 2026-05-08 Categories: eess.SY, cs.LG arXiv: https://arxiv.org/abs/2605.07052v1 We generalize Jan Willems' behavioral approach to a class of discrete-time nonlinear systems in a vector-valued reproducing kernel Hilbert space (RKHS). Apart from linear time-invariant systems, this class covers nonlinear systems modeled by Volterra series and their autoregressive variants, as well as systems admitting Hammerstein-type state-space realizations. We apply the proposed framework to the problem of data-driven modeling of such systems, i.e., when simulation or control objectives for an unknown system are carried out without an explicit system identification step. To that end, we link the behavioral approach to two data-driven modeling methods in a vector-valued RKHS: (1) minimum-norm interpolation and (2) subspace identification. --- ### NSMQ Riddles: A Benchmark of Scientific and Mathematical Riddles for Quizzing Large Language Models Authors: George Boateng, Naafi Ibrahim, Samuel John, Philemon Badu, Patrick Agyeman-Budu, Jonathan Mensah, Kevin Yeboah, William Edor, Andrew Mensa-Onumah, Nana Yeboah, Victor Wumbor-Apin Kumbol Published: 2026-05-08 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.07051v1 Large Language Models (LLMs) have shown good performance on various science educational benchmarks, demonstrating their potential for use in science and mathematics education. Yet, LLMs tend to be evaluated on science and mathematical educational datasets from the Western world, with an underrepresentation of datasets from the Global South. Furthermore, they tend to have multiple-choice answer options that are trivial to evaluate. In this work, we present NSMQ Riddles, a novel benchmark of Scientific and Mathematical Riddles from Ghana's National Science and Maths Quiz (NSMQ) competition to evaluate LLMs. The NSMQ is an annual live TV competition for senior secondary school students in Ghana that brings together the smartest high school students in Ghana who compete in teams of 2 by answering questions in biology, chemistry, physics, and math over five rounds and five stages until a winning team is crowned for that year. NSMQ Riddles consists of 11 years of riddle questions (n=1.8K) from the 5th round, with each riddle containing a minimum of 3 clues. Students compete to be the first to guess the answer on any of the clues, with earlier clues being vague and also fetching more points. The answers are usually a number, word, or short phrase, allowing for automatic evaluation. We evaluated state-of-the-art models: closed (GPT-5.4, Gemini 3.1 Pro, Claude Opus 4.6) and open models (Kimi-K2.5, DeepSeek-V3.1, GPT-OSS-120B) with high and low reasoning settings. Our evaluation shows that the dataset is challenging even for state-of-the-art LLMs, which performed worse than the best student contestants. This work contributes a novel and challenging benchmark for scientific and mathematical reasoning from the Global South towards enabling a true global benchmarking of LLMs' capabilities for science and mathematics education. --- ### Self-Organising Digital Circuits Authors: Marcello Barylli, Gabriel Béna, Alexander Mordvintsev, Eleni Nisioti, Sebastian Risi Published: 2026-05-07 Categories: cs.AI, cs.NE arXiv: https://arxiv.org/abs/2608.02606v2 Fault tolerance in classical computing has traditionally relied on static strategies like hardware redundancy and error-correcting codes. Biological systems, in contrast, exhibit adaptive plasticity, maintaining function through dynamic re-organisation around damage. Inspired by this principle, we introduce Self-Organising Digital Circuits, framing functional logic generation and maintenance as a meta-learning problem on graphs. Our architecture employs a topology-masked Transformer that configures the Lookup Tables (LUT) of a circuit's Boolean gates. Extending the pattern-generation paradigm of Neural Cellular Automata (NCA), it navigates the degenerate Boolean search space to satisfy a computational task, rather than regenerating a fixed target state. We demonstrate that it can self-assemble functional circuits from scratch and rapidly re-route logic around permanent, previously unseen hardware faults. For soft errors, the policy achieves near-perfect recovery (>99.99\% accuracy) from damage sizes far exceeding training conditions. We further observe generalisation across circuit scales: accuracy improves on graphs substantially wider than those seen during training. This work bridges the principles of biological self-organisation with the practical domain of digital hardware. --- ### Self-Organising Digital Circuits Authors: Marcello Barylli, Gabriel Béna, Alexander Mordvintsev, Eleni Nisioti, Sebastian Risi Published: 2026-05-07 Categories: cs.AI, cs.NE arXiv: https://arxiv.org/abs/2608.02606v1 Fault tolerance in classical computing has traditionally relied on static strategies like hardware redundancy and error-correcting codes. Biological systems, in contrast, exhibit adaptive plasticity, maintaining function through dynamic re-organisation around damage. Inspired by this principle, we introduce Self-Organising Digital Circuits, framing functional logic generation and maintenance as a meta-learning problem on graphs. Our architecture employs a topology-masked Transformer that configures the Lookup Tables (LUT) of a circuit's Boolean gates. Extending the pattern-generation paradigm of Neural Cellular Automata (NCA), it navigates the degenerate Boolean search space to satisfy a computational task, rather than regenerating a fixed target state. We demonstrate that it can self-assemble functional circuits from scratch and rapidly re-route logic around permanent, previously unseen hardware faults. For soft errors, the policy achieves near-perfect recovery (>99.99\% accuracy) from damage sizes far exceeding training conditions. We further observe generalisation across circuit scales: accuracy improves on graphs substantially wider than those seen during training. This work bridges the principles of biological self-organisation with the practical domain of digital hardware. --- ### It Takes 8 Tokens: Weak-to-Strong Off-Policy RL via Auxiliary Branches Authors: Dayu Wang, Jiaye Yang, Weikang Li, Jiahui Liang, Liwei Qian, Xin Pei, Jizhou Huang Published: 2026-05-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.16205v1 Reinforcement learning with verifiable rewards has emerged as a standard approach for enhancing reasoning in large language models, which typically optimizes the policy by contrasting multiple self generated rollouts. However, we identify a critical support limited bottleneck in this paradigm: on challenging reasoning tasks, the target model's samples often exhibit semantic redundancy, converging into the same erroneous "reasoning basins" that offer negligible reward contrast for policy updates. In this paper, we propose to overcome this limitation through a weak to strong learning paradigm, where a policy's exploration is informed by a weaker but computationally efficient auxiliary model. We introduce W2SPO, an off policy RL method that injects short auxiliary segments often as brief as 8 tokens into intermediate target model trajectories and the target model then completes the reasoning path from these diverted states. Policy updates are restricted to these short inserted segments based on final verifiable rewards. Empirically, W2SPO achieves superior performance among evaluated 4B scale models on mathematical reasoning benchmarks, outperforming evaluated post trained baselines. Compared with vanilla GRPO under the same sampling budget, W2SPO improves Pass@1 from 62.3% to 64.2% while achieving a 3.55 times training speedup. These results suggest that weak auxiliary branches can induce stronger target reasoning policies by expanding local exploration support. --- ### Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RL Authors: Darshan Deshpande Published: 2026-05-07 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2607.16204v1 Recent growth in reinforcement learning (RL) has surfaced a need for diverse, specialized training environments. Hand-curated environments with fixed task and reward difficulties become ineffective signals as model performance improves, and sparse rewards over long horizons induce mode collapse on specific workflows or tool structures. World models that simulate environment states have matched pure rollout performance, making them promising for scaling diversity on-demand. However, autoregressive (AR) world models suffer from a left-to-right bias preventing conditioning on globally interdependent state anchors such as tool schemas, prior turns, and expected outcomes. We (i) formalize text-based world modeling as a steerable transition-dynamics problem decomposed into initial state, task context, tool schemas, domain rules, and steering directives, and (ii) curate 239,403 grounded state-action trajectories spanning nine open-source environments and twelve frontier model families. We compare AR LMs and masked diffusion language models (MDLMs), showing MDLMs, via bidirectional anchor-aware denoising, achieve better coherence, groundedness, and empirically validated rollout diversity than LLMs over 4x their parameter size, at comparable inference latency. We introduce a plug-and-play GRPO training framework with deterministic state checks, and perform zero-shot transfer ablations on three OOD environments (ScienceWorld, ALFWorld, AppWorld) across three 1.2B-7B agent backbones (LFM2.5, Qwen3, Mistral), achieving up to 47% absolute gains over baselines without environment-specific fine-tuning. We further conduct behavioral analysis of failure modes under adversarial scenarios and human evaluation on realism, outcome correctness, and training utility. We open-source our work to encourage research in this direction. --- ### MAPS: Modeling Co-Existing Subjective Perspectives and Shared Meaning in Multi-Agent Cognitive Dialogue Authors: Molood Arman, Clément Bonnafous Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.14110v1 Human dialogue involves more than exchanging information; it also expresses beliefs, emotions, and subjective cognitive styles. Yet current AI dialogue systems often enforce semantic uniformity, sacrificing diversity and interpretability. We present MAPS (Multi-Agent Perspective Spaces), a novel framework that models dialogue between cognitively distinct agents through domain-weighted profiles, dynamic GRU-based memory, and interpretable token-level attention. MAPS enables agents to maintain individualized reasoning while progressively converging on shared meaning. Evaluations on EmpatheticDialogues, TopicalChat, and MultiWOZ show that MAPS supports semantic alignment without collapsing subjectivity. Our results demonstrate a path toward cognitively grounded, interpretable dialogue systems that balance expressiveness and coherence. --- ### ReasonLab: A Controlled and Auditable Evaluation of Prompting Techniques for Multiple-Choice QA Authors: Inder Preet, Shuxin Lin, Dhaval Patel Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.14109v2 Probing the capabilities of Large Language Models (LLMs) and building robust solutions for Multiple-Choice Question Answering (MCQA) remain central challenges in natural language understanding. Furthermore, the rapid proliferation of LLMs has created the implicit assumption that more sophisticated prompting techniques yield better performance. Several studies claim such gains, but report them under differing models, prompt wordings and answer-extraction rules, so the gains cannot be attributed to the technique alone. We address this gap with ReasonLab, an evaluation framework in which the prompting technique is a first-class experimental variable alongside the model and the dataset, and which retains every generation for inspection. Using ReasonLab we conduct a controlled study of 8 prompting techniques across 10 MCQA datasets, 27 model configurations and 480,927 evaluations at temperature 0. We find that the prompting technique is a minor determinant of accuracy: on configurations without a reasoning budget the reasoning triggers improve on direct prompting by only 3.92 to 4.69 pp and are indistinguishable from one another, and on configurations with reasoning enabled no technique differs by more than 0.51 pp. Self-Generate is the only technique with a consistent effect, a reduction of 2.95 pp. We further investigate three phenomena: (1) the comparison of models on a common set of datasets, where model size does not predict accuracy, (2) the trade-offs across thinking budgets, where enabling reasoning is worth up to 12.74 pp whereas an eightfold budget increase adds only 0.48 to 2.10 pp, and (3) the variation in dataset difficulty, with 60% of benchmarks below 70% accuracy and a 43.9 pp spread from easiest to hardest. These results suggest that, for MCQA, the prompting technique is a minor lever compared with enabling model reasoning, and that substantial headroom remains. --- ### Simplicity Paradox: Debunking myths about prompting and datasets for LLM evaluation Authors: Inder Preet, Shuxin Lin, Dhaval Patel Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.14109v1 Probing the capabilities of Large Language Models (LLMs) and building robust solutions for Multiple-Choice Question Answering (MCQA) remain central challenges in natural language understanding. Furthermore, the rapid proliferation of LLMs has created the implicit assumption that more sophisticated prompting techniques yield better performance. Several studies claim better performance with more sophisticated prompting techniques, but do not provide a comprehensive evaluation. We address this gap through a comprehensive empirical study of 8 prompting techniques across 10 multiple-choice question answering (MCQA) datasets, encompassing 27 model configurations and roughly 4,300 unique questions evaluated more than 430,000 times. Our findings reveal a striking paradox that baseline prompting consistently outperforms complex reasoning techniques on various benchmarks. Only minimal expert and inductive role framing (CoT-Expert and CoT-Inductive) yields a small but statistically significant $\sim$3 percentage-point (pp) gain over baseline whereas every other elaborate technique we tested matches or under-performs it, often by large margins (up to 31~pp for Self-Analogical). We further investigate three critical phenomena: (1) the unexpected victory of Qwen3-30B-A3B-Thinking-2507 in Elo ratings, (2) the performance-efficiency trade-offs across model variants with different thinking budgets, revealing model-dependent optimal configurations, and (3) the substantial variation in dataset difficulty, with 60% of benchmarks below 70% accuracy and a 47.5~pp spread from easiest to hardest, indicating considerable room for model improvement. These results suggest that the LLM evaluation community may be overcomplicating prompt engineering and that substantial performance gaps remain across diverse benchmarks, offering opportunities for genuine model improvements rather than prompt optimization. --- ### Eta Given Delta: Defining LLM Tool Efficiency With Marginal Tool Utility Authors: Nyx Iskandar, Perla Gamez Published: 2026-05-07 Categories: cs.CL, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2607.14108v2 This paper introduces tool efficiency, a new quantitative metric to evaluate the rate of useful tool calls in an LLM agent trajectory. To ensure that tool efficiency is well-defined, we also introduce marginal tool utility, a new quantitative metric defined per tool call indicating whether a tool is useful or whether it can be safely removed from the tool suite without affecting accuracy while increasing tool efficiency; in this paper, we determine the sign of marginal tool utility for each tool call in a trajectory using LLM-as-a-Judge. While much prior work has been done to develop techniques that improve tool use by LLMs and design evaluation methods measuring efficiency indirectly using accuracy as a proxy, our work is centered on measuring efficiency directly via the quantitative metric proposed in this paper in post hoc trajectory analyses. It is our intention that this work contributes to the frontier of LLM evaluation research as a springboard for future benchmark designs and agent harness engineering (specifically with regards to creating lean tool suites) that optimize for metrics that complement but are distinct from accuracy. --- ### Eta Given Delta: Defining LLM Tool Efficiency With Marginal Tool Utility Authors: Nyx Iskandar Published: 2026-05-07 Categories: cs.CL, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2607.14108v1 This paper introduces tool efficiency, a new quantitative metric to evaluate the rate of useful tool calls in an LLM agent trajectory. To ensure that tool efficiency is well-defined, we also introduce marginal tool utility, a new quantitative metric defined per tool call indicating whether a tool is useful or whether it can be safely removed from the tool suite without affecting accuracy while increasing tool efficiency; in this paper, we determine the sign of marginal tool utility for each tool call in a trajectory using LLM-as-a-Judge. While much prior work has been done to develop techniques that improve tool use by LLMs and design evaluation methods measuring efficiency indirectly using accuracy as a proxy, our work is centered on measuring efficiency directly via the quantitative metric proposed in this paper in post hoc trajectory analyses. It is our intention that this work contributes to the frontier of LLM evaluation research as a springboard for future benchmark designs and agent harness engineering (specifically with regards to creating lean tool suites) that optimize for metrics that complement but are distinct from accuracy. --- ### Polestar: Drift-Aware Cache Calibration and Token Commitment for Efficient Inference of Diffusion LLMs Authors: Mingyu Lee, Akshat Ramachandran, Souvik Kundu, Tushar Krishna Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.14107v1 The inference efficiency of diffusion large language models (dLLMs) is constrained by two challenges: bidirectional attention precludes efficient KV-cache reuse, while increasing decoding parallelism with static confidence thresholds can compromise generation quality. We observe that both challenges arise from a shared phenomenon: as tokens are decoded, their contextual integration through bidirectional attention causes token representations to drift (evolve) across decoding steps. This insight motivates Polestar, a training-free inference framework that uses token representation drift as a unified signal to jointly address both challenges. Polestar comprises two components: Polestar-Cache, which identifies stale KV-cache positions via drift and performs sparse KV-cache refreshes to enable efficient reuse, and Polestar-Commit, which detects sharp drift events to reliably identify commit-ready tokens. Across mathematics and coding benchmarks on several dLLM families, Polestar sets a new state of the art on the accuracy-throughput Pareto frontier, achieving up to 10.73% accuracy improvement, up to 3.7x higher throughput, and high decoding parallelism of 3.67 tokens per forward pass over existing baselines. --- ### Token Time Continuous Diffusion for Language Modeling Authors: Parikshit Bansal, Sujay Sanghavi Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.14106v1 In this paper we introduce token time continuous diffusion (TTCD), a new diffusion language model which (a) operates in continuous space, deterministically mapping Gaussian noise to a final token canvas with no further sampling, and crucially (b) incorporates a new notion of per-token times, with some tokens proceeding from noise to token at a faster rate than others. Continuous space modeling helps TTCD avoid the parallel sampling of multiple tokens, which is a key source of inaccuracy at high speedups for models that iterate purely in discrete space. The notion of per-token times helps TTCD to better model conditional generation, allows for more sure tokens to proceed at a faster rate, and allows for differentiated inter-token influences during refinement. TTCD outperforms discrete models at high speedups. We train a 160M parameter TTCD model on OpenWebText, and then self-distill it; we find that at high speedups we are comparable in unconditional generation quality, and outperform in conditional generation, several existing models of similar size trained, on the same data, and self-distilled. We achieve similar gains in Sudoku solving as well. --- ### Automatically Evolving Prompt Guidelines for Task-Specific Optimization Authors: Cedric Richter, Salah Ghamizi, Mike Papadakis Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.14105v1 For Large Language Models to reliably answer user queries, users must clearly specify requirements, context, and constraints. In practice, however, user queries are often underspecified, forcing models to infer unstated assumptions that may misalign with the actual user intent. Existing prompt engineering guidelines aim to mitigate this issue, they are typically generic and task-agnostic, limiting their practical utility. Additionally, existing guidelines are formed manually and in a non-systematic way. To this end, we study prompt guideline optimization: the problem of automatically generating task-specific guidelines that help write better-specified prompts for a given task and model. Our key observation is that existing (completed) task examples (aka reference answers) often implicitly encode the missing information required to complete underspecified queries, including behavioral constraints, contextual assumptions, and evaluation criteria. We therefore propose AGOPS, an automatic approach that evolves task-specific guidelines via an optimization scheme that involves a prompt LLM writer, a solver LLM and prompt evolution, which maximize downstream effectiveness on a set of examples (user queries with reference answers). At inference time, our guidelines help users write well-specified prompts, boosting the effectiveness of LLMs. We show across mathematical reasoning, medical question answering, and coding tasks, that prompt underspecification leads to major drops (up to 95.3%) in downstream task performance (compared to well-specified prompts) and, perhaps more importantly, that this drop can hardly be recovered by existing prompt optimization techniques. Users following AGOPS guidelines can regain this loss (increasing performance between 15.5 to 81.7% on average) consistently across all benchmarks. --- ### I'm Sorry, but I Can't Help with Braille: Revealing Accessibility Failures in State-of-the-Art LLMs Authors: Abdullah Abdullah Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.11893v1 Large Language Models (LLMs) perform strongly on many language tasks, but their capability in structurally constrained, accessibility-critical modalities such as Braille remains unclear. We evaluate state-of-the-art LLMs on bidirectional Korean-Braille translation using a human-annotated dataset. Despite expectations that multilingual, instruction-tuned models can generalize to Braille via text representations, we find consistently poor, unstable outputs and substantial disagreement with human judgments. These results point to missing Braille-aware tokenization and weak alignment between Korean and Braille patterns. In contrast, supervised fine-tuning of a small model (T5-small) on the same data yields large and stable gains over zero-shot and prompted LLM baselines across standard metrics (SacreBLEU, ChrF++, CER, BLEU, ROUGE-L, METEOR, CIDEr). Our findings reveal a systematic limitation of current LLMs and demonstrate the effectiveness of modest task-specific supervision. --- ### G-SHARE: A Guideline-Based Structured Reasoning Framework for Human-Factor Event Diagnosis Authors: Xingyu Xiao, Mao Du, Jiejuan Tong, Jingang Liang, Haitao Wang Published: 2026-05-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.11892v1 Human-factor event diagnosis is essential for learning from operational events in nuclear power plants, yet its quality depends strongly on expert interpretation of narrative reports and guideline-based reasoning.Existing data-driven or one-shot large language model approaches often lack structured reasoning, have limited alignment with formal diagnostic guidelines, and may generate logically inconsistent conclusions. To address this issue, this study proposes G-SHARE, a guideline-based structured reasoning framework that operationalizes the CNNP nine-step human-factor event diagnosis guideline into a multi-stage diagnostic pipeline.The framework consists of evidence extraction, stepwise diagnostic reasoning, and post-hoc consistency repair, enabling explicit use of report evidence, intermediate rationale generation, and logical validation of diagnostic outputs. A dataset of real human-factor event reports was constructed from Chinese nuclear industry sources, and a gold-standard subset annotated by domain experts was used for evaluation. Results show that G-SHARE substantially outperforms one-shot prompting and traditional machine learning baselines, with the strongest version achieving the best overall accuracy and macro-F1. Ablation results further indicate that structured reasoning and consistency enforcement are critical to robust diagnosis, especially under weak prompting conditions. The findings demonstrate the value of transforming expert diagnostic guidelines into auditable reasoning workflows, providing a practical pathway for intelligent human-factor analysis in safety-critical industries. --- ### SWIFT: Spatio-temporal Wavelet Integrated Forecasting Framework for Workload Traces Authors: Zeyuan Ding, Lingfeng Zheng, Dian Ding, Guangtao Xue Published: 2026-05-07 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2607.02524v1 Accurate cloud workload forecasting is pivotal for efficient resource management but remains challenging as workloads are highly volatile and prone to sudden bursts. Although wavelets preserve temporal locality, rigid fixed bases struggle with complex patterns and isolated processing neglects critical spatial dependencies. To address this, we propose SWIFT, a pure convolutional framework designed for high-efficiency workload forecasting. We introduce a Learnable Cascaded Wavelet Path that reformulates the traditional fixed wavelet bases into adaptive convolutional operators, enabling precise, data-driven feature peeling. Complementing this, our Multivariate Interaction Module sequentially models inter-variable spatial and intra-variable feature interactions to stabilize and refine noisy workload states. Extensive experiments demonstrate that SWIFT achieves SOTA accuracy with linear O(L) complexity, reducing prediction error by up to 31.04% while cutting latency by 79.74%. --- ### ADEPT: An Entropy-Driven Dual-Strategy Agent for Interactive Video Retrieval Authors: Ke Chen, Shengyuan Han, Yongfeng Huang, Yujin Zhu, Jingwei Xiong, Liang Xu, Jundong Liu Published: 2026-05-07 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2606.28326v1 This research aims to solve the challenge of video retrieval from massive datasets, caused by ambiguous user queries. Prevailing single-round retrieval paradigms face a performance bottleneck, as they lack effective feedback mechanisms to handle complex search intentions. The root cause is the "Intent-Query Gap", where users' intent cannot be captured by a simple text query. To solve this, we propose the ADEPT framework: a training-free agent that pioneers an entropy-driven decision engine to efficiently guide dialogue by dynamically selecting between ASK and REFINE strategies. Experiments on two challenging datasets demonstrate that ADEPT significantly outperforms all non-interactive, heuristic, and Video-LLM baselines. The core contribution of this work is an efficient and interpretable entropy-driven interactive strategy that sets a new performance benchmark for the field of interactive video retrieval. --- ### Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs Authors: Fahd Seddik, Fatemeh Fard Published: 2026-05-07 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2606.27378v1 We introduce an axiomatic evaluation framework for latent thought representations in LLMs, comprising metrics that are independent of downstream benchmark scores and reveal representational failures that benchmark accuracy masks. Existing evaluations conflate representation quality with model capacity. Therefore, failures cannot be attributed to the representation rather than to the model that processes it. We formalize four functional axioms (Causality, Minimality, Separability, and Stability) and define a quantitative measure for each, computed directly on the representation independently of downstream accuracy. We audit open-weight LLMs across 23 reasoning tasks (e.g., Spatial Reasoning, Factual QA). We find that no candidate satisfies all four axioms simultaneously, that the representations distinguish task type reliably but cannot distinguish between two questions within the same task, and that the representations encode little information beyond what is already present in the input embedding. The failure is consistent across dense, reasoning-distilled, and RL-trained model families, indicating that the gap is structural rather than a property of model size or training procedure. --- ### Examining Human-Like Behaviors in LLMs: A Multi-Dimensional Analysis of Model Behaviors, User Factors, and System Prompts Authors: Sunnie S. Y. Kim, Margit Bowler, Leon A Gatys Published: 2026-05-07 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2606.18258v1 Large language models (LLMs) exhibit a wide range of human-like behaviors, from expressing thoughts and emotions, to engaging in relationship-building with users, to refusing requests and maintaining boundaries. Despite their prevalence, researchers and practitioners lack methods and empirical insights to make informed decisions about when and what types of human-like behaviors LLMs should exhibit. To fill this gap, we present a multi-dimensional analysis of the prevalence, potential effects, and controllability of these behaviors using LLM-as-a-judge and human evaluation. Across 21,000 multi-turn conversations from four widely used models (gpt-4o, gpt-4.1-mini, claude-sonnet-4.6, gemini-2.5-flash), we find that human-like behaviors are pervasive but vary across models and user factors (conversation goals and user profiles). In terms of perceived appropriateness, human evaluators judged self-referential and relationship-building behaviors as less appropriate from LLMs than from humans, but boundary-maintaining behaviors more appropriate from LLMs than from humans. Finally, we show that system prompting can control these behaviors, though it requires careful evaluation to avoid unintended effects. We discuss the implications of our findings and provide recommendations for responsible LLM design and evaluation. --- ### Towards Distributed Inference of LLMs on a P2P Network Authors: Shabari S Nair, Krishanu Saini Published: 2026-05-07 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2606.17059v1 Prefix caching can reduce LLM inference latency by reusing KV caches across requests with shared prompts, but cluster-scale reuse is challenging because caches are partitioned across nodes. We propose a decentralized, prefix-cache-aware routing scheme for peer-to-peer LLM serving. Each node maintains a local radix tree of its own cached prefixes and asynchronously refreshed estimates of peer caches using periodic anti-entropy. Requests are routed to the node with the longest estimated prefix match, without centralized coordination or KV-cache transfer. Stale metadata only causes cache misses, not incorrect outputs, making weak consistency sufficient for correctness. Evaluation on simulated MMLU workloads show that decentralized routing improves latency under low communication delay and skewed prefix distributions, while high network latency and affinity-induced hotspots limit its benefits. --- ### The AI Legal Specialist: A Juridically Autonomous Professional Profile for AI Governance Authors: Nicola Fabiano Published: 2026-05-07 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2606.12415v1 The rapid global expansion of artificial intelligence regulation has generated, across multiple jurisdictions, a demand for legal expertise dedicated to AI that the market has addressed in a fragmented manner. Data protection officers extend their remit beyond data protection law; privacy lawyers reposition themselves toward AI; compliance officers add AI chapters to their existing manuals. This paper argues that none of these adaptive responses adequately covers the professional space opened by the emerging global AI regulatory landscape, of which the EU Artificial Intelligence Act (Regulation (EU) 2024/1689) is the most comprehensive instance, alongside the Council of Europe Framework Convention on AI, the United States executive and sectoral framework, and analogous initiatives in the United Kingdom, Canada, Brazil, China, Japan, Singapore, and beyond. A distinct professional profile is required: the AI Legal Specialist, conceived as a jurist -- understood broadly to encompass any professional with advanced legal training -- operating at the intersection of legal interpretation and AI governance. The profile is juridically autonomous: it derives its existence from the structure of regulatory obligations generated wherever AI is subject to substantive regulation, rather than from any technical standard or the extension of adjacent roles. The paper provides a juridically grounded definition of the profile, argues for its autonomy from adjacent figures and international standards, proposes a reference competence architecture aligned with the European e-Competence Framework (e-CF, EN 16234-1) as a methodological choice, and articulates the conditions for its operational measurement through key performance indicators. The contribution is intended as a foundation for international standardization of the profile and as a reference for practice, curricula, and adoption across jurisdictions. --- ### PyCC.id: A package for hypothesis-driven equation discovery with structural identifiability Authors: Federico J. Gonzalez Published: 2026-05-07 Categories: cs.LG, eess.SP arXiv: https://arxiv.org/abs/2606.05191v1 Data-driven equation discovery is fundamentally an inverse problem that seeks to infer the governing differential equations of a system directly from time-series measurements. A known issue is the ill-conditioned nature of the inverse problem, which frequently produces multiple mathematical models that fit the data similarly well. One path to address this issue is by incorporating known hypotheses and constraints into the training phase beforehand. While this approach effectively reduces the search space, it still results in multiple candidate models, forcing practitioners to rely on post-hoc manual filtering based on their own domain expertise. A recent approach incorporates structural `skeletons' inspired by characteristic curves (CCs), defining a hypothesis-driven methodology. In this methodology, practitioners define a skeleton, which is associated with a family of ordinary differential equations (ODEs), and then add their hypotheses and priors based on their domain knowledge to refine the obtained model iteratively. An important advantage of this approach is that some skeletons have demonstrable structural identifiability properties, which are useful for checking whether the skeleton is correct or should be discarded. Furthermore, this formalism enables the use of multiple equation discovery paradigms due to its modularity (such as neural networks, symbolic regression, and sparse regression). In this work, we present the Python library PyCC, which condenses these efforts into a flexible tool that allows researchers and engineers to seamlessly define their skeletons and hypotheses to discover ODEs from time-dependent data. --- ### Grokers: Bottom-Up Inductive Comprehension and Write-Time Intelligence over Typed Knowledge Graphs Authors: Gregory Magarshak Published: 2026-05-07 Categories: cs.AI, cs.CL, cs.DB, cs.IR arXiv: https://arxiv.org/abs/2606.00050v1 We present Grokers, an architecture for building persistent, structured comprehension of typed knowledge graphs through bottom-up inductive traversal of dependency subgraphs. Unlike retrieval-augmented generation (RAG), which pays full comprehension cost at every query, Grokers pushes intelligence to write time: autonomous Groker agents analyze nodes in a typed stream graph, extract structured attributes via governed language model (LM) calls, and inductively compose that understanding upward through dependency relations, writing enriched typed attributes that serve all future queries at zero additional LM cost. We prove three formal properties: (1) the Byte-Identity Theorem, establishing that context blocks assembled from a transactionally-maintained denormalization index are byte-identical across LM turns between semantic changes, enabling KV-cache hit rates approaching 100%; (2) the Accumulation Monotonicity Theorem, establishing that the fraction of interactions resolved without LM calls is non-decreasing in the number of completed interactions under a governed wisdom library growth protocol; and (3) the Dual-Traversal Ordering Theorem, establishing that top-down generation and bottom-up comprehension are the unique correct traversal orderings for their respective tasks over a dependency DAG, and that their composition closes into a complete generation-comprehension cycle. We further present a deterministic alternative to embedding-based semantic search, with a synonym caching protocol whose LM fallback rate converges to zero for finite-vocabulary domains. A reference implementation is provided in the open-source Qbix / Safebox / Safebots stack. --- ### A Simple State Space Model Excels at Multivariate Time Series Classification Authors: Hassan Saadatmand, Geoffrey I. Webb, Hamid Rezatofighi, Mahsa Salehi Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.27406v1 Structured state space models (SSMs) have recently emerged as a promising foundation for sequence modeling, with Mamba-based architectures demonstrating strong performance through input-dependent state transitions, albeit at considerable complexity. However, their application to time-series classification (TSC) has been largely limited to Mamba-style architectures, leaving the broader SSM design space underexplored. We present the first systematic study spanning diagonal SSMs (S4D) and input-dependent SSMs (Mamba family) on large-scale TSC benchmarks, asking whether such complexity is necessary for top performance. Our results reveal a surprising finding: S4D consistently outperforms Mamba-based variants in both accuracy and efficiency, challenging the assumption that increased complexity translates to meaningful gains in TSC. Building on this, we introduce MS4, lightweight modifications to S4D via a linear input projection and channel-mixing mechanism, and MS4N, a normalized variant that stabilizes state dynamics with negligible overhead. Evaluated on 59 datasets across MONSTER (up to 60 million samples, 50K timesteps, 82 classes) and the UEA benchmark, against 15 baselines, MS4 and MS4N consistently outperform Mamba-based models while remaining more efficient, and MS4N matches or surpasses competing deep learning models that are roughly 2x and 10x larger in parameters. These results position lightweight structured SSMs as a compelling alternative to scaling complexity for TSC. --- ### Stop Comparing LLM Agents Without Disclosing the Harness Authors: Yunbei Zhang, Janet Wang, Yingqiang Ge, Weijie Xu, Jihun Hamm, Chandan K. Reddy Published: 2026-05-07 Categories: cs.AI, cs.SE arXiv: https://arxiv.org/abs/2605.23950v1 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. --- ### Features have life history. And we should care Authors: Philipp Stecher, Sandro Radovanović, Vlasta Sikimić, Reinhard Kahle Published: 2026-05-07 Categories: q-bio.NC, cs.AI arXiv: https://arxiv.org/abs/2605.18789v1 Features in language models have life history: they emerge, persist, and die during training, yet the importance of that history remains largely unexplored. We find evidence of a persistent representational backbone, which we identify in Pythia-160M and -410M as the carrier scaffold: ${\sim}50$ sparse features with stable life histories, around which the model's representational structure organises. It has four properties. \emph{(i)}~\emph{It assembles early:} features emerge, die, and reorganise ${\sim}40\!\times$ faster in the first $1\%$ of training than afterwards, and the scaffold is already largely fixed by then. \emph{(ii)}~\emph{It is load-bearing:} joint cross-layer ablation identifies the carriers as far more load-bearing than any count-matched non-scaffold population, a gap invisible to per-firing single-feature methods. \emph{(iii)}~\emph{Function precedes direction:} which features will become carriers is already predictable from training-onset firing patterns alone, correctly distinguishing future carriers from non-carriers in $4$ of $5$ cases, before the geometry has settled. \emph{(iv)}~\emph{It seeds subsequent development:} by the end of training, scaffold carriers have recruited $64\%$ of all active features into the scaffold hierarchy. Life history is consistent with a two-phase account of training: selection appears to largely determine the scaffold in the first $1\%$; the remaining $99\%$ appears to calibrate geometry around a substrate already set. --- ### Orth-Dion: Eliminating Geometric Mismatch in Distributed Low-Rank Spectral Optimization Authors: Tatsuhiro Nakamori, Laura Gomezjurado Gonzalez, Ganesh Talluri, Ansh Tiwari, Hideyuki Kawashima, Ioannis Mitliagkas, Guillaume Rabusseau, Hiroki Naganuma Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.16341v1 Low-rank gradient compression reduces communication in distributed training by representing updates with rank-$r$ factors. Dion is a recent method that approximates Muon, a spectral optimizer that orthogonalizes momentum, using one step of power iteration followed by column normalization (rescaling each column of the right factor to unit length). This makes it compatible with fully sharded data parallel training, but it converges more slowly than full-rank spectral methods. We show that this gap is geometric: column normalization does not yield the rank-$r$ polar factor that Muon implicitly targets, so the resulting direction violates the dual-norm constraint of the low-rank spectral geometry, and the rate picks up an extra factor of $\sqrt{r}$ even though the low-rank approximation of the gradient itself is accurate. The same mismatch enters the smoothness term and the error-feedback recursion in the analysis, which has a knock-on effect on empirical performance. We propose Orth-Dion, which replaces column normalization with QR orthogonalization of the right factor. Under non-Euclidean smoothness, with $L_r$ the curvature constant along rank-$r$ directions, Orth-Dion attains rate $O(\sqrt{L_r/T})$, matching exact spectral methods at the same per-step communication cost as Dion. The proof removes the bounded-drift assumption common in prior error-feedback analyses via a self-consistent fixed-point argument, and uses a time-averaged contraction that only requires the error sequence to contract on average rather than at every step. Experiments on large-scale language model pre-training validate the predicted $\sqrt{r}$ scaling and show that Orth-Dion closes the convergence gap to Muon at Dion's communication cost. --- ### Preference Instability in Reward Models: Detection and Mitigation via Sparse Autoencoders Authors: Shunchang Liu, Xin Chen, Belen Martin Urcelay, Francesco Croce Published: 2026-05-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.16339v1 Preference learning in large language models relies on reward models as proxies for human judgment. However, these models frequently exhibit preference instability, producing contradictory preference assignments in response to subtle, meaning-preserving input variations. We analyze this instability at the representation level under three semantic-preserving perturbation types: paraphrasing, pattern injection, and backdoor triggers. We attribute this instability to over-reliance on predictive yet brittle features, which we term unstable features, and isolate them via Sparse Autoencoders (SAEs) in a sparse latent space where benign and perturbed inputs activate distinctly separable patterns. Building on this separability, we propose two SAE-based instability mitigation strategies: SAE Feature Steering, which identifies and suppresses anomalously activated features at inference, and SAE Residual Correction, which learns adaptive adjustments over SAE features to restore correct preferences. Our methods substantially reduce incorrect preference assignments on harmlessness and hallucination benchmarks while preserving benign performance and general utility on other tasks, without retraining the reward model. Our code and data are available in \url{https://github.com/shunchang-liu/pisa}. --- ### Vidya: An AI-Driven Modular Pipeline for Archival Automation and Semantic Metadata Enrichment Authors: Cloter Migliorini Filho, Julia Graciela Machado, Edson Armando Silva, Marcella Scoczynski Published: 2026-05-07 Categories: cs.DL, cs.CL arXiv: https://arxiv.org/abs/2605.16338v1 The large-scale digitization of historical archives has created a paradox: "dark data"-digital objects lacking metadata for retrieval. Manual archival description is slow and expensive, limiting discovery and reuse. We propose Vidya, a modular pipeline that orchestrates Large Language Models (LLMs) and FOSS tools to automate semantic enrichment and archival ingestion at scale. Vidya constrains generations using YAML-defined ontologies and Pydantic validation, producing deterministic, structured JSON outputs from probabilistic models. Developed at Laboratory for Digital Humanities and Innovation (LAMUHDI) of the State University of Ponta Grossa (UEPG), Vidya applies Maker principles and open-source practices to enable low-cost deployment in memory institutions using modest hardware. We compare LLM performance and present a cost-benefit analysis showing major gains, reducing processing time from decades to days while complying with NOBRADE and ISAD(G). --- ### Detecting Verbatim LLM Copy-Paste in Homework Authors: Aizierjiang Aiersilan Published: 2026-05-07 Categories: cs.CR, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2605.16336v1 Large language models (LLMs) have made fluent essay writing, code drafting, and quiz answering instantly available to students at every level, from secondary school through graduate study. Many educators do not object to LLM use \emph{per~se}; what they need to detect is the case in which a student pastes the assignment prompt into a chatbot and submits the model's reply verbatim, without engaging with the work. Existing post-hoc AI-text detectors remain unreliable and have been shown to penalise non-native English writers, while output-side watermarks require cooperation from the model provider. We propose an alternative that the educator controls directly: an input-side watermark in which an invisible instruction is embedded inside the visible assignment prompt itself. An LLM that ingests the prompt verbatim quietly reads the hidden instruction and writes a tell-tale signature into its reply, exposing the copy-and-paste pathway specifically. We describe SteganoPrompt, a single-page, zero-dependency web tool that encodes an arbitrary printable-ASCII payload into the deprecated Unicode Tags block (\texttt{U+E0000}--\texttt{U+E007F}). The encoded string is visually identical to the original, survives common copy-paste channels (Word, Google Docs, PDF, Markdown, Slack, e-mail, the major learning-management systems), and is reliably tokenized by frontier models. We evaluate compliance across seven LLM families and a representative set of educational content channels. The work is informed by my experience as a graduate teaching assistant for an undergraduate software engineering course at the George Washington University. The tool is released under the MIT licence at \url{https://ezharjan.github.io/SteganoPrompt/}. --- ### Decentralized Conformal Novelty Detection via Quantized Model Exchange Authors: Kyle Loh, Yu Xiang Published: 2026-05-07 Categories: stat.ML, cs.IT, cs.LG, eess.SP, stat.ME arXiv: https://arxiv.org/abs/2605.08263v1 This work studies decentralized novelty detection with global false discovery rate (FDR) control across heterogeneous composite null distributions, without sharing the raw data due to privacy and bandwidth considerations. We propose a framework based on the exchange of quantized surrogate models, allowing independent agents to share low-precision representations of locally learned non-conformity score functions. We prove that evaluating data against these quantized composite scores preserves conditional exchangeability, providing rigorous finite-sample guarantees for global FDR control. Empirical studies on synthetic datasets confirm our theoretical results, demonstrating that the proposed approach maintains competitive statistical power while drastically reducing the communication cost. --- ### SLayerGen: a Crystal Generative Model for all Space and Layer Groups Authors: Rees Chang, Andrew Novick, Ryan P Adams, Elif Ertekin Published: 2026-05-07 Categories: cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2605.08262v1 Crystal generative models have shown rapid progress for accelerating the discovery of bulk, periodic materials. However, many material systems such as 2D superconductors, thin film semiconductors, and catalytic surfaces are diperiodic, i.e., aperiodic along one of the lattice directions. These systems are invariant under the layer groups, which are known to influence materials properties yet not considered by existing models. In this paper, we propose SLayerGen, a generative model that produces crystals constrained to be invariant to any space or layer group. SLayerGen consists of coarse-to-fine discrete autoregressive lattice generation; transformer-based autoregressive sampling of Wyckoff positions, elements, and numbers of symmetrically unique atoms; and space or layer group equivariant diffusion of atomic coordinates. For the diffusion component, we corrected an inconsistency in the loss from prior work arising from hexagonal groups being non-orthogonal in fractional coordinates. To facilitate progress in generative modeling of diperiodic materials, we assembled and filtered datasets of monolayers and bilayers, propose relevant evaluation metrics, and developed novel representations for layer group symmetries. For de novo generation of diperiodic materials, SLayerGen achieves consistent performance gains over bulk crystal generative models and is competitive when training jointly on bulk and diperiodic materials. --- ### Computer Use at the Edge of the Statistical Precipice Authors: Pierluca D'Oro, Sneha Silwal, William Wong, Yuxuan Sun, Fanyi Xiao, Manchen Wang, Eric Gan, Allen Bolourchi, Joseph Tighe Published: 2026-05-07 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2605.08261v1 Evaluating Computer Use Agents (CUAs) on interactive environments is fraught with methodological pitfalls that the field has yet to systematically address. We show that a 1MB replay script that blindly executes a recorded action sequence without ever observing the screen outperforms frontier models on prominent static benchmarks, and prove that its expected success rate is exactly equal to the source agent's pass@k in deterministic environments. We trace this and other failures to two root causes: non-principled environment design (static, unsandboxed, or unreliably verified environments) and non-principled evaluation methodology (naive aggregation and misuse of pass@k for stateful UI interactions). To address the first, we propose PRISM, five design principles for CUA environments (privileged verification, realistic environments, integrity-checked configurations, sandboxed execution, and multifactorial variability) and instantiate them in DigiWorld, a benchmark of 15 realistic sandboxed mobile applications able to evaluate agents in over 3.2 million verified unique configurations. To address the second, we develop an aggregation framework pairing Wilson score intervals with hierarchical bootstrap, producing confidence intervals that correctly account for the nested structure of CUA benchmarks, as we empirically demonstrate. All together, we show that principled environment design and rigorous evaluation methodology are not optional refinements but prerequisites for meaningful CUA research. --- ### Research on Security Enhancement Methods for Adversarial Robust Large Language Model Intelligent Agents for Medical Decision-Making Tasks Authors: Saisai Hu Published: 2026-05-07 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.08257v1 Motivated by the challenge to improve the adversarial robustness, security, and trust of medical decision making intelligent agents, this study develops a full-link security enhancement framework, which describes "input risk perception - medical evidence constraint - knowledge consistency verification - decision confidence reweighting - security output control - adversarial feedback update." We propose ARSM-Agent and define a weighted joint objective consisting of decision accuracy loss, adversarial robustness loss, safety refusal loss, and knowledge consistency loss, with weights of 0.3, 0.3, 0.2, and 0.2, respectively. The whole medical decision formulation is implemented by multi-module collaborative linkage. We verify that the algorithm is more efficient than four baselines, including LLM-Agent, Retrieval-Agent, Filter-Agent, and Adv-Train-Agent. Under semantic perturbation, prompt injection, drug-name confusion, and false-evidence attacks, ARSM-Agent reduces the overall attack success rate to 8.7% and achieves a knowledge consistency score of 0.91. Ablation experiments quantify each module's contribution: removing risk perception, evidence retrieval, consistency verification, and confidence reweighting reduces accuracy by 6.7%, 9.1%, 7.6%, and 4.4%, respectively, and increases attack success rate by 13.8%, 11.1%, 8.6%, and 6.9%. The proposed approach addresses key security issues of medical decision making intelligent agents, obtains secure decision making in challenging scenarios, and provides reliable intelligent support for medical decision-making intelligent agents. --- ### Can LLMs Predict Polymer Physics Just by Reading Synthesis and Processing Prose? Authors: Yuchu Liu, Rui Zhu, Jingwei Xiong, Haixu Tang Published: 2026-05-07 Categories: cs.LG, cond-mat.mtrl-sci, cs.AI arXiv: https://arxiv.org/abs/2605.08255v1 Can large language models predict physical and mechanical polymer properties simply by reading unstructured scientific prose? Polymer performance is rarely determined by chemical structure alone; identical nominal polymers can exhibit drastically different behaviors depending on their synthesis route, processing history, morphology, and testing conditions. Yet, state-of-the-art polymer property models typically rely on structure-only representations -- such as SMILES or molecular graphs -- which strip away this vital experimental context. In this work, we introduce \textbf{PolyLM}, a natural-language-only, process- and condition-aware framework that predicts materials performance directly from full-text literature. By circumventing structural inputs entirely, PolyLM preserves the nuanced, unstructured descriptions of synthesis and processing reported by domain scientists. To train this framework, we curated an unprecedented, literature-scale dataset encompassing 185,000 scientific papers and over 276,400 unique polymer samples across 22 physical, mechanical, and thermal properties. We fine-tuned a massive 9-billion-parameter language model (Qwen3.5-9B) using Low-Rank Adaptation (LoRA) and task-level uncertainty weighting. Evaluated on 68,283 held-out observations, the model achieves remarkably high predictive accuracy, establishing new state-of-the-art benchmarks for complex properties. Across the 22 diverse targets, the model achieves a median $R^2$ of 0.74, with predictions for key thermal, mechanical, and physicochemical properties frequently surpassing an $R^2$ of 0.80. These results unequivocally demonstrate that natural language is a powerful, highly scalable interface for realistic materials performance prediction. --- ### HyperTransport: Amortized Conditioning of T2I Generative Models Authors: Valentino Maiorca, Eleonora Gualdoni, Xavier Suau, Marco Cuturi, Luca Zappella, Pau Rodríguez Published: 2026-05-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.08254v1 As foundation models grow in capability, the ability to efficiently and reliably control their behavior becomes critical. Fine-tuning these models can be costly, and while prompting can be practical for controllability, it remains fragile due to models' high sensitivity to exact prompt wording and structure. This brittleness has driven interest in activation steering techniques that offer more stable and predictable control over model behavior. However, existing activation steering methods require per-concept optimization, which makes them ill-suited to deployment scenarios where the concept set is large, evolving, or only specified at request time: each new concept incurs at least minutes of optimization on the target model. We propose HyperTransport, a hypernetwork framework that amortizes this cost by mapping embeddings from a pretrained encoder (CLIP in our instantiation) directly to intervention parameters, trained end-to-end using an optimal transport loss. Once trained, HyperTransport produces each new intervention in a single hypernetwork forward pass, 3600-7000x faster than per-concept fitting. On concepts unseen during training, it matches the strongest per-concept baselines at inducing the target concept. By decoupling concept representation from intervention prediction, HyperTransport combines three capabilities that no existing approach offers as a set: amortized steering for open-ended concept sets, continuous interpretable strength control, and cross-modal conditioning where reference images can directly steer text-based generation. We validate HyperTransport on DMD2 and Nitro-1-PixArt across 167 held-out test concepts via CLIP-based metrics, a VLM-as-a-judge evaluation, and a user study. In pairwise comparisons, both human and VLM judges prefer HyperTransport over prompting ~2x as often. --- ### Path-Coupled Bellman Flows for Distributional Reinforcement Learning Authors: Boyang Xu, Qing Zou, Siqin Yang, Hao Yan Published: 2026-05-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.08253v2 Distributional reinforcement learning (DRL) models the full return distribution, but existing finite-support or quantile-based methods rely on projections, while recent flow-based approaches can suffer from \emph{boundary mismatch} at the flow source or from \emph{high-variance} bootstrapping when current and successor noises are independent. We propose Path-Coupled Bellman Flows (PCBF), a continuous-time DRL method that learns return distributions with flow matching using \textbf{source-consistent Bellman-coupled paths}: the current path starts from the required base prior at $t{=}0$, reaches the Bellman target at $t{=}1$, and maintains a pathwise affine relation to the successor flow at intermediate times (without requiring time-$t$ marginals to satisfy a distributional Bellman fixed point for all $t$). PCBF couples current and successor return flows through shared base noise and uses a $λ$-parameterized control-variate target: $λ{=}0$ recovers an unbiased sample Bellman target, while $λ{>}0$ trades controlled bias for variance reduction. Experiments on analytically tractable MRPs, OGBench, and D4RL show improved distributional fidelity and training stability, and competitive offline RL performance. --- ### Multimodal Emotion Recognition via Causal-Diffusion Bridge (Affect-Diff) Authors: Ankit Sanjyal Published: 2026-05-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.08252v1 Multimodal emotion recognition on CMU-MOSEI faces an extreme imbalance as Happy accounts for 65.9% of samples while three Ekman categories collectively represent under 7%, causing standard fusion models to maximize accuracy by ignoring minority emotions entirely. We present Affect-Diff, a Causal-Diffusion Bridge that addresses this through three jointly trained mechanisms: a NOTEARS-learned causal graph that re-weights modality contributions before fusion, a beta-VAE bottleneck for regularized latent compression, and a stop-gradiented 1D DDPM prior that structures the latent space against majority-class collapse. On 3,292 aligned CMU-MOSEI samples, Affect-Diff achieves validation balanced accuracy 0.384, an 18% relative improvement over the strongest baseline (TETFN: 0.324), while all evaluated baselines produce zero F1 on Fear, Disgust, and Surprise. Ablation studies confirm independent, non-redundant contributions from the diffusion prior (-24% without it) and causal graph (-13%). Notably, only the deterministic-encoder variant detects all six emotion classes, revealing KL regularization strength as a direct lever for minority-class sensitivity. --- ### Why Do DiT Editors Drift? Plug-and-Play Low Frequency Alignment in VAE Latent Space Authors: Xiaoce Wang, Sifan Zhou, Kaifei Wang, Leli Xu, Xuerui Qiu, Tao He, Ming Li Published: 2026-05-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.08250v1 Recent advances in diffusion transformers (DiTs) have enabled promising single-turn image editing capabilities. However, multi-turn editing often leads to progressive semantic drift and quality degradation.In this work, we study this problem from a latent-space frequency perspective by decomposing the editing process into two functional components: VAE and DiT. Through systematic analysis in the VAE latent space, we uncover that the DiT introduces dominant low-frequency drift that accumulates as semantic misalignment across editing rounds, while the VAE contributes comparatively stable reconstruction bias.Based on this insight, we propose VAE-LFA (Low Frequency Alignment), a training-free, plug-and-play method that performs alignment in VAE latent space. VAE-LFA decomposes latent discrepancies across editing rounds via low-pass filtering, and aligns low-frequency statistics to an exponential moving average of previous rounds, effectively suppressing accumulated semantic drift while preserving high-frequency details.Our method requires no retraining, ground-truth priors, or access to diffusion parameters, making it applicable to both white-box and black-box DiT editors. For white-box models, VAE-LFA is seamlessly integrated into the editing pipeline by eliminating redundant VAE round trips; for black-box models, it operates via an off-the-shelf VAE to perform inter-round latent alignment.Extensive experiments demonstrate that VAE-LFA improves semantic consistency and visual fidelity across diverse multi-turn editing scenarios, including both controlled and in-the-wild images. --- ### Dimensional Coactivation for Representational Consistency in Frozen Vision Foundation Models Authors: Izaldein Al-Zyoud Abdulmotaleb El Saddik Published: 2026-05-07 Categories: cs.CV, eess.IV, eess.SP arXiv: https://arxiv.org/abs/2605.08249v1 Frozen vision foundation models do not merely extract features; they organize images through a learned coordinate system. We ask whether that coordinate system remains internally coherent within a single input. This leads to Representational Consistency: the study of whether a frozen foundation model represents one sample coherently across its semantic subregions. We introduce Dimensional Coactivation (DCA), a per-dimension instrument for measuring this coherence. DCA compares semantic regions by asking whether the same feature dimensions coactivate across them. Unlike classical similarity measures, it deliberately avoids centering, L2 normalization, and full Gram coupling. These operations are useful when comparing different models or distributions, but they are mismatched to the intra-sample setting, where the coordinate system is fixed and raw magnitude carries signal. Deepfake detection provides a natural validation task. Synthetic faces may reproduce plausible eyes, noses, and mouths while breaking the representational structure that links those regions in real faces. Using frozen DINOv3 features, DCA exposes this break: an eyes-mouth-nose fingerprint achieves 0.9106 AUC on CelebDF-v2 and 0.9289 on DFD under FF++ c23 cross-dataset transfer. The design is also sharply validated by ablation: reintroducing centering collapses CelebDF-v2 AUC to 0.459, L2 normalization reduces it to 0.862, and cross-dimension coupling reduces it to 0.478. Finally, replacing DINOv3 with FaRL collapses CelebDF-v2 AUC to 0.582. DCA therefore depends on a stable per-dimension coordinate system, not on region extraction alone. These results position DCA as an instrument for measuring intra-sample representational coherence in frozen foundation models, with deepfake detection as the first validation task. --- ### LLM Translation of Compiler Intermediate Representation Authors: Andrea Valenzuela Ramirez, Cristian Gutierrez-Gomez, Marta Barroso, Dario Garcia-Gasulla, Sara Royuela Published: 2026-05-07 Categories: cs.PL, cs.AI arXiv: https://arxiv.org/abs/2605.08247v1 GCC and LLVM underpin much of modern software infrastructure, relying on distinct Intermediate Representations (IRs) to drive optimizations and code generation. However, the semantic and structural differences between these IRs create significant barriers for cross-toolchain interaction, limiting the reuse of compiler frontends, backends, and optimization pipelines across programming languages and compilation ecosystems. Traditional rule-based translators have attempted to bridge this gap, but their complexity and maintenance cost have hindered practical adoption. In this context, Large Language Models (LLMs) appear to be an emerging technology that offers a data-driven alternative, capable of learning complex mappings between heterogeneous compiler IRs directly from sufficiently representative examples. To explore this approach, this paper presents IRIS-14B, a 14-billion-parameter transformer model fine-tuned to translate GIMPLE (as emitted by GCC) to LLVM IR (as emitted by LLVM). The model is trained on paired IRs extracted from C sources and evaluated on the GIMPLE-to-LLVM IR transformation applied to IRs derived from real-world C code and competitive programming problems. To the best of our knowledge, IRIS-14B is the first model trained explicitly for IR-to-IR translation. It outperforms the accuracy of widely used models, including the largest state-of-the-art open models available today, ranging from 13 to 1,000 billion parameters, by up to 44 percentage points. The proposed transformation supports the integration of LLMs as complementary components within hybrid neuro-symbolic compiler architectures, where models such as IRIS-14B act as interoperability layers enabling cross-toolchain workflows without modifying existing compiler passes, while traditional compiler infrastructure continues to perform deterministic compilation and optimization. --- ### Smart Railway Obstruction Detection System using IoT and Computer Vision Authors: Pravin Kumar, Mritunjay Shall Peelam, Ramakant Kumar, Sanjay Kumar, Vinay Chamola Published: 2026-05-07 Categories: cs.CV, cs.CR, cs.LG arXiv: https://arxiv.org/abs/2605.08246v1 Railway track intrusions pose a critical safety challenge for Indian Railways, encompassing wildlife incursions and deliberate malicious obstructions. The December 2025 collision in Assam, in which seven elephants were killed by the Rajdhani Express, underscores the urgency of effective real-time detection. Existing solutions such as the optical fiber-based Gajraj system suffer from prohibitive costs (\$1000/km) and high false alarm rates, limiting deployment to only 20 of India's 101 elephant corridors. This paper proposes NETRA, a cost-effective, internet-independent intrusion detection system deployed on Raspberry Pi Zero W and Raspberry Pi 4 edge platforms. NETRA employs probabilistic sensor fusion integrating a PIR motion sensor and an HC-SR04 ultrasonic distance sensor with a tunable threshold (tau_c = 0.65), enabling event-driven camera activation that reduces unnecessary visual processing by 52%. Upon confirmed intrusion, edge-AI classification using MobileNet-SSD (Pi Zero) or YOLOv5 ONNX (Pi 4) identifies threats including humans, large animals, and track obstructions. Confirmed threats are transmitted via LoRa (868 MHz) to alert the locomotive driver within 2.4 seconds end-to-end. Experimental evaluation across 113 motion events demonstrated 95% detection accuracy with zero false alarms through probabilistic fusion, compared to 85% for binary methods. Raspberry Pi 4 with YOLOv5 achieved 83.5% elephant F1-score, a 5.6x improvement over Pi Zero's heuristic approach (14.8%). LoRa communication achieved 100% packet delivery across 1-2 km in field trials. NETRA reduces deployment cost by 75% (\$247/km vs \$1000/km for Gajraj) while providing unified detection of both wildlife and obstruction threats. --- ### When Language Overwrites Vision: Over-Alignment and Geometric Debiasing in Vision-Language Models Authors: Harshvardhan Saini, Samyak Jha, Yiming Tang, Dianbo Liu Published: 2026-05-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.08245v4 Vision-Language Models (VLMs) increasingly power high-stakes applications, from medical imaging to autonomous systems, yet they routinely hallucinate, confidently describing content not present in the input. We investigate the root causes of these failure modes with a mechanistic analysis focusing on the decoder-based VLMs. We trace these failure modes to a geometric over-alignment: to bridge the modality gap required by attention mechanisms, decoder-based VLMs over-align visual embeddings with the text manifold, injecting a statistical linguistic bias that systematically overshadows fine-grained visual evidence. While prior work either aggressively closes this gap or suppresses hallucinations through expensive black-box decoding strategies, none addresses the underlying geometric cause. We provide the first quantitative characterization of this over-alignment, demonstrating that linguistic bias concentrates in the top principal components of a universal, dataset-agnostic text subspace. Building on this insight, we propose two complementary remedies: a training-free inference strategy and a bias-aware fine-tuning paradigm, both of which explicitly project out this subspace from visual representations. Our methods significantly reduce hallucinations across POPE, CHAIR, and AMBER benchmarks, and improve CLAIR scores on long-form captioning tasks, with the training-free variant adding no computational overhead over the base model. --- ### GPU-Accelerated Synthesis of Mixed-Boolean Arithmetic: Beyond Caching Authors: Gabriel Bathie, Baptiste Mouillon, Nathanaël Fijalkow Published: 2026-05-07 Categories: cs.PL, cs.DC, cs.LG arXiv: https://arxiv.org/abs/2605.08243v1 Synthesizing Mixed-Boolean Arithmetic (MBA) expressions from input-output examples is central to program deobfuscation and also useful for compiler optimization, reverse engineering, and cryptanalysis. Existing MBA synthesizers are typically CPU-based and scale poorly on large specifications or complex targets. Recent GPU-accelerated synthesis methods achieve large speedups in qualitative settings, but they depend on caching observationally equivalent candidates; this strategy breaks down for MBA because candidate outputs are quantitative bitvectors and the behavioral space is enormous. We present SIMBA (Synthesis of Mixed-Boolean Arithmetic), a GPU-accelerated MBA synthesizer built around cache-free bottom-up enumeration. SIMBA avoids language caches entirely and uses a GPU-oriented enumeration design that keeps work local and highly parallel. In experiments, SIMBA is substantially faster than prior MBA synthesis tools, handles larger specifications, and reaches expression sizes that existing methods fail to solve. These results establish cache-free GPU synthesis as a practical and scalable approach for quantitative domains, and identify it as a strong alternative to cache-centric designs. --- ### An Explainable Unsupervised-to-Supervised Machine Learning Framework for Dietary Pattern Discovery Using UK National Dietary Survey Data Authors: Wing Yi Yu, Chun Yin Chiu Published: 2026-05-07 Categories: q-bio.QM, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.08242v1 Clinical dietary assessment can generate detailed but high-dimensional nutrient and food-group information that is difficult to translate quickly into counselling priorities. This paper proposes an explainable unsupervised-to-supervised machine learning framework for discovering, reproducing and interpreting dietary patterns using public UK National Diet and Nutrition Survey data. Adult participants aged 19 years and above from NDNS Years 12-15 were represented using 25 energy-adjusted nutrient and food-group features. K-means, Gaussian Mixture Models and Agglomerative Clustering were compared across k = 2-8, with stability and dietetic interpretability used alongside internal validation metrics. The selected K-means k = 4 solution identified four interpretable dietary patterns: high fat/meat and sodium, higher fibre fruit-vegetable micronutrient, high free-sugar snacks and sugary drinks, and dairy/cereal calcium-rich saturated-fat. A supervised surrogate classifier reproduced held-out cluster membership with high test performance (macro-F1 = 0.963), but was interpreted only as an explanatory surrogate rather than as an independent clinical prediction model. SHAP analysis linked predictions to dietetically meaningful drivers, suggesting potential value for dietitian-in-the-loop assessment, counselling prioritisation and follow-up monitoring. --- ### TinySSL: Distilled Self-Supervised Pretraining for Sub-Megabyte MCU Models Authors: Bibin Wilson Published: 2026-05-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.08241v1 Self-supervised learning (SSL) has transformed representation learning for large models, yet remains unexplored for microcontroller (MCU)-class models with fewer than 500K parameters. We identify three obstacles at this scale -- projection head dominance, representation bottleneck, and augmentation sensitivity -- and propose Capacity-Aware Distilled Self-Supervised Learning (CA-DSSL), a teacher-guided framework that overcomes them without labels or text supervision. CA-DSSL combines asymmetric distillation from a frozen DINO ViT-S/16 teacher, multi-scale feature distillation for spatial representations, and a progressive augmentation curriculum. On a MobileNetV2-0.35 backbone (396K parameters) pretrained on CIFAR-100, CA-DSSL reaches 62.7 0.5% linear-probe accuracy (3-seed mean) -- surpassing SimCLR-Tiny by 18 pp, matching SEED (61.7%) with 10 fewer projection parameters (426K vs. 3.15M), and reaching 94.0% of a supervised upper bound. Standard SSL methods (BYOL-Tiny, DINO-Tiny) collapse entirely at this scale. On Pascal VOC detection, CA-DSSL achieves 2.3 the mAP of random initialization and +3 pp over SEED, though SimCLR-Tiny matches CA-DSSL on detection mAP. The deployed backbone occupies 378 KB (INT8) with no inference overhead from pretraining. Preliminary ImageNet-100 experiments reveal that CA-DSSL's advantage is specific to small-data regimes; scaling to ImageNet-1K is discussed as future work. --- ### Resource-Aware Evolutionary Neural Architecture Search for Cardiac MRI Segmentation Authors: Farhana Yasmin, Mahade Hasan, Haipeng Liu, Amjad Ali, Ghulam Muhammad, Yu Xue Published: 2026-05-07 Categories: cs.CV, cs.AI, cs.ET, cs.LG arXiv: https://arxiv.org/abs/2605.08238v1 Cardiac magnetic resonance (CMR) segmentation underpins quantitative assessment of ventricular structure and function, yet reliable delineation remains difficult due to low tissue contrast, fuzzy boundaries, and inter scan variability. We present CardiacNAS, an evolutionary neural architecture search (NAS) framework that couples a UNet like supernet with a cardiac aware search space spanning depth width, kernel size, filter size, attention, fusion, activation, dropout, and residual scaling. The search is explicitly resource aware, jointly optimizing dice similarity coefficient (DSC) and 95th percentile Hausdorff distance (HD95) versus model size and floating point operations (FLOPs) under fixed compute budgets. Candidate architectures are instantiated from the supernet, trained with proxy budgets, and evolved through crossover, mutation, and elitist selection. We evaluate on the ACDC dataset and compare against six state of the art methods, using qualitative comparisons, learning curve analyses, and design factor correlation studies. The resulting model attains 93.22% average DSC and 4.73 mm HD95 with 3.58M parameters and 14.56 GFLOPs, demonstrating a favorable accuracy efficiency trade off. Analyses indicate that searched attention and fusion choices, together with residual scaling, contribute to improved boundary fidelity and stability. CardiacNAS offers a principled, resource aware approach to deployable CMR segmentation with transparent reporting of architectural complexity and compute budgets. ---