# Research Papers — 2026-05 · bundle page 309 of 326 > Full-text markdown of papers #15401–15450 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/309 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/308.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/310.md) ### Geometric and Spectral Alignment for Deep Neural Network I Authors: Ziran Liu, Wei Wang, Jinhao Wang, Pengcheng Wang, Xinyi Sui, Cihan Ruan, Nam Ling, Wei Jiang Published: 2026-05-04 Categories: cs.LG, math.DG arXiv: https://arxiv.org/abs/2605.02108v1 Deep residual architectures are modeled as products of near-identity Jacobians. This paper proves deterministic quotient-geometric estimates for singular spectra of Frobenius-normalized layer factors, emphasizing a normalized top-radial Cartan coordinate and fitted power-law chart. Full-rank factors are mapped from $\mathrm{GL}(d)$ to the positive cone by $A\mapsto A^\top A$, then to ordered eigenvalue data. Under Frobenius normalization, exact power-law spectra form a trace-normalized Cartan orbit. This orbit is a Gibbs family on ranks, a Fisher information line, and a Bures--Wasserstein curve with line element $d/4$ times Fisher information. The main rigidity theorem is a slack-aware margin inequality: interface radial amplitude, non-backtracking slack, and signed residual variation control displacement of the fitted Cartan coordinate. In the exact-chart zero-slack case, a depth-$L$ budget gives exponent drift of order $(\log M)/L$; generally, slack and residual increments augment the bound. We separate scalar top-radial from full-Cartan spectral control, which also needs Bures/Hellinger residual variation. We prove approximate-power-law and metric-chart versions, converse lower bounds, Fisher--KL/Bures action estimates, and near-identity expansions for normalized residual chains. Near-identity results verify transport budgets; chart quality remains measurable. Effective rank is a spectral-energy quantile, giving finite-width power-law tail bounds and robust rank-window transition estimates. Empirical static-weight exponent profiles serve as diagnostics; full verification also requires interface budgets, slacks, and residuals for the same operator chain. --- ### The Dynamic Gist-Based Memory Model (DGMM): A Memory-Centric Architecture for Artificial Intelligence Authors: Terry Dorsey, Kevin Huggins Published: 2026-05-04 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.02106v1 Contemporary artificial intelligence systems achieve strong performance through large-scale parameterization, retrieval augmentation, and training on extensive static corpora. Despite these advances, they continue to face limitations in persistent memory, temporal grounding, provenance, and interpretability. These challenges are especially pronounced in large language models, where experience is encoded implicitly in fixed parameters, limiting the ability to preserve, inspect, and reinterpret past interactions over time. This paper establishes a memory-centric architectural foundation for artificial intelligence in which experience is represented explicitly and persistently to support temporal grounding, provenance, and interpretability. It proposes an alternative to parameter-centric approaches by treating memory as a first-class, structured substrate for reasoning. We introduce the Dynamic Gist-Based Memory Model (DGMM), an architecture in which experience is represented as an evolving, graph-structured episodic-semantic memory. DGMM encodes experience as interconnected conceptual structures grounded in time, source, and interaction context, and defines selective, cue-conditioned recall as the mechanism for constructing working memory. A formal schema and architectural invariants are provided based on additive memory growth and recall-conditioned interpretation. The results specify properties of DGMM, including episodic persistence, locality of cue-conditioned surprise, and contextual variability without structural modification of stored memory. DGMM provides a coherent architectural theory in which memory is explicit and persistent, supporting evolving interpretation without retraining and enabling interpretable, context-aware, and temporally grounded AI systems. --- ### Sharpness-Aware Pretraining Mitigates Catastrophic Forgetting Authors: Ishaan Watts, Catherine Li, Sachin Goyal, Jacob Mitchell Springer, Aditi Raghunathan Published: 2026-05-04 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2605.02105v1 Pretraining optimizers are tuned to produce the strongest possible base model, on the assumption that a stronger starting point yields a stronger model after subsequent changes like post-training and quantization. This overlooks the geometry of the base model which controls how much of the base model's capabilities survive subsequent parameter updates. We study three pretraining optimization approaches that bias optimization toward flatter minima: Sharpness-Aware Minimization (SAM), large learning rates, and shortened learning rate annealing periods. Across model sizes ranging from 20M to 150M parameters, we find that these interventions consistently improve downstream performance after post-training on five common datasets with up to 80% less forgetting. These principles hold at scale: a short SAM mid-training phase applied to an existing OLMo-2-1B checkpoint reduces forgetting by 31% after MetaMath post-training and by 40% after 4-bit quantization. --- ### Auxiliary uncertainty signals for LLM-assisted systematic review screening: a benchmark across eight Cohen drug-class reviews Authors: Arya Rahgozar, Pouria Mortezaagha Published: 2026-05-03 Categories: cs.CL, cs.DL, cs.IR, cs.LG arXiv: https://arxiv.org/abs/2608.14551v1 Large language models (LLMs) are increasingly used for title-abstract screening in systematic reviews, but their decisions lack calibrated uncertainty. We show that an auxiliary BERT+GCN classifier supplies a structured uncertainty signal that improves LLM screening efficiency, and we identify the prompt-delivery strategy that maximises the benefit-to-cost ratio. We evaluate five LLM prompt-delivery conditions on eight drug-class datasets from the Cohen (2006) benchmark using 3 seeds x 5-fold stratified cross-validation (600 fold-level results). A BERT+GCN model trained per fold classifies each test paper as INCLUDE, EXCLUDE, or MAYBE via two spectral tests (algebraic radical and categorical paradox). Conditions vary information content (none / label / full scores), selectivity (all papers vs. MAYBE only), and timing (proactive vs. reactive two-pass). A cross-model pilot against gpt-4.1-mini on three datasets tests cross-generation transfer. Three findings: (i) Full-context delivery yields significant gains in F1 (+0.011, paired Wilcoxon p=0.008) and WSS@95 (+0.050, p=0.039) at a 1.28x token-cost premium, while preserving recall. (ii) MAYBE-only routing is Pareto-optimal: highest mean recall (0.92) and AUC-ROC (0.54) at only 1.05x baseline cost -- one sixth of full-context overhead. (iii) The two-pass design escalates 22.2% +/- 8.8% of records yet never revises its decision (0% flip rate across all datasets and folds), giving decisive evidence that current instruction-tuned LLMs cannot self-triage. The cross-model pilot shows an identical +0.8% recall uplift for both LLM generations. A per-paper ablation across 20,796 observations shows the dual paradox test reduces empirically to a one-line logit-gap criterion. We release the full pipeline; the 600-run experiment replays in under one hour from cached LLM responses. --- ### From Frame-Level Recognition to Event-Level Confirmation: Repair Traces and Runtime Failure Analysis of Public-Space Gesture Interaction Authors: M. Meng, Yansong Zhang Published: 2026-05-03 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.21601v1 Public-space gesture interaction is often evaluated as a frame-level recognition problem, but deployed systems expose a different failure boundary. In scenic kiosks, exhibition halls, and service terminals, users experience whether an intended action becomes a stable interaction event, not whether individual hand-landmark frames are correct. We call this the recognition-to-interaction gap. This paper analyzes 8 engineering repair records from a scenic-area interactive kiosk project, covering 4 gesture tasks: two-hand bowing, single-hand fist shaking, two-hand catching control, and knowledge-graph node hovering. From these traces, we extract 20 failure instances and organize them into six non-exclusive working failure classes: model-output degeneration, temporal mismatch, geometric-scale instability, coordinate-rendering mismatch, runtime lifecycle failure, and feedback synchronization and recovery failure. We further organize recurring repair mechanisms into an event-level runtime abstraction between the hand-landmark model and the interaction task. The contribution is deliberately bounded: a deployment-grounded failure taxonomy, an event-confirmation runtime abstraction, and case-study findings. We do not claim a new recognition model, large-scale user evaluation, or quantified accuracy gains. --- ### Securing Multimodal AI through Internal Information Decomposition Authors: Jehyeok Yeon, Hyeonjeong Ha, Qiusi Zhan, Heng Ji Published: 2026-05-03 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.21600v1 Multimodal large language models introduce attack surfaces absent in unimodal systems: adversaries can distribute malicious intent across modalities to evade unimodal safeguards. This motivates using cross-modal consistency as a detection signal rather than inspecting each modality in isolation. Our key observation is that benign inputs induce compatible predictive behavior from text-only and vision-only reasoning that stabilizes when fused, whereas adversarial manipulation disrupts this consistency, causing abnormal multimodal behavior. Existing defenses that examine raw inputs or outputs overlook this internal fusion process, rendering them brittle and computationally expensive. We propose FlowGuard, a lightweight inference-time framework that detects harmful inputs by monitoring internal multimodal consistency. Unlike approaches that rely on scalar confidence metrics, FlowGuard derives FlowVectors inspired by Partial Information Decomposition that quantify cross-modal redundancy, synergy, and modality-specific dominance, capturing whether fused multimodal predictions remain aligned with unimodal semantic evidence. In a one-class classification problem trained solely on benign data, FlowGuard reduces Attack Success Rates from >90% to <15% on unseen attacks, with <3% utility loss and up to a 6 times latency reduction. Our results demonstrate that monitoring cross-modal consistency offers an efficient and effective defense for multimodal reasoning. --- ### Decoupled Attention Fusion: Accelerating RAG with Efficient KV Cache Reuse Authors: Xiabao Wu, Wentao Liu, Yongchao Liu, Jiajun Zheng Published: 2026-05-03 Categories: cs.PF, cs.AI arXiv: https://arxiv.org/abs/2607.21599v1 Retrieval-Augmented Generation (RAG) effectively mitigates hallucinations in Large Language Models (LLMs) but suffers from prohibitive Time-To-First-Token (TTFT) latency in long-context scenarios. Reusing pre-computed document KV caches addresses this but introduces a distribution mismatch, where offline caches lack the inter-document attention patterns required for coherent reasoning. CacheBlend reduces recomputation via selective attention, but suffers severe accuracy degradation at longer contexts. To address these challenges, we propose Decoupled Attention Fusion (DAF), a framework that maintains high accuracy while significantly reducing recomputation overhead. DAF decouples the attention process into three integrated stages: important-token self-attention to restore missing inter-document attention, question-document self-attention for standard inference, and a state fusion that concatenates their outputs to synthesize the final hidden states. By decoupling these operations into dense patterns, DAF is natively compatible with Flash-Attention kernels, maximizing hardware utilization without requiring complex attention masks. Experiments show that DAF delivers up to 2 times speedup over CacheBlend and 5.6 times over full recomputation with vLLM on long-context benchmarks, without sacrificing accuracy. --- ### FixItFlow: Automated Troubleshooting Guide Generation from Cloud Incidents Authors: Srihari Unnikrishnan, Jaskaran Singh Walia, Drishti Goel, Supriyo Ghosh Published: 2026-05-03 Categories: cs.CL, cs.AI, cs.LG, cs.SE arXiv: https://arxiv.org/abs/2607.13035v1 Cloud services experience frequent incidents that require rapid diagnosis and resolution. Troubleshooting guides help engineers respond consistently, but creating them manually is labor-intensive, resulting in incomplete coverage and outdated documentation. We present FixItFlow, an automated system that generates troubleshooting guides from historical incident data using large language models. The system extracts diagnostic patterns from engineer actions, synthesizes structured guides with verified commands, and enforces strict validation to prevent fabricated content. In our evaluation with 26 engineers, generated guides achieved 61.5\% positive ratings for clarity and demonstrated a 2.3x reduction in mitigation time for incidents with associated guides. These results indicate that automated guide generation can improve incident response while reducing documentation burden on engineering teams. --- ### HALO: Hybrid Adaptive Latent Reasoning for Language Models Authors: Micah Zhang Published: 2026-05-03 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2607.08775v1 We study how to improve a frozen pretrained language model with a small amount of adaptive extra computation. A simple approach is to add additional refinement steps on top of the backbone hidden states, but fixed extra refinement can be wasteful: a one-step refinement head may be too weak, while forcing a second full-sequence refinement step everywhere can increase compute without improving transfer. We introduce HALO, a hybrid adaptive latent-refinement method that combines a coarse refinement stage with selective second-stage latent refinement on a subset of tokens chosen by token scoring and monotonic token halting. On the main public benchmark comparison built from MMLU-Pro and GPQA-Diamond, HALO achieves the best overall average among the paper-facing methods, outperforming the frozen backbone, fixed-1, and fixed-2. Internal analysis further shows that HALO reaches nearly the same token-accuracy level as fixed-2 while using fewer average applied refine steps than fixed-1 and far fewer than fixed-2. These results suggest that the key advantage is not simply more refinement, but a better allocation of refinement: HALO achieves the strongest paper-facing result while also using less measured controller compute than either fixed baseline. --- ### Benchmarking KV-Cache Optimizations across Task Quality and System Performance for Long-Context Serving Authors: Nikita Agrawal, Ruben Mayer Published: 2026-05-03 Categories: cs.CL, cs.AI, cs.DC arXiv: https://arxiv.org/abs/2607.05399v1 Large language model serving is increasingly limited by KV-cache growth under long-context workloads, yet existing KV-cache compression techniques are difficult to compare because they were evaluated on different models, tasks, budgets, and serving stacks. This paper presents a workload-aware benchmark of representative KV-cache optimization mechanisms spanning quantization, pruning, and merging, including KIVI, TurboQuant, SnapKV, and CaM, evaluated on LongBench-style multi-document QA, single-document QA, few-shot learning, and summarization workloads using Llama-3.1-8B-Instruct and Mistral-7B-Instruct-v0.3. The benchmark measures task quality, mean output throughput, mean time-to-first-token, and realized compression ratio across context-length buckets. The results show that the compression ratio alone is a poor predictor of end-to-end performance. KIVI4 provides the most stable quality across models, SnapKV delivers the strongest long-context throughput, and CaM yields large gains on selected QA workloads but exhibits substantial workload sensitivity in both quality and realized compression ratio. These findings motivate workload-aware selection of KV-cache mechanisms rather than one-size-fits-all compression and provide deployment guidance for long-context serving systems. --- ### How Personas Can Influence Agents to Play Split or Steal Authors: Carlos Leon, Alexandre Rodrigues, Pedro Gamito, Thomas D. Parsons Published: 2026-05-03 Categories: cs.CL, cs.CY arXiv: https://arxiv.org/abs/2607.05398v1 Personas are often employed to guide large language model agents, yet their effectiveness in shaping strategic behavior in social dilemma settings remains uncertain. To address this, we examined the impact of persona prompts in an iterated Split or Steal game where persona-driven agents interacted with a Virtual Human (VH) controlled by a fixed prompt. Agents were instantiated from four open models (Ministral 3:3b, phi4:14b, Gemma3:12b, and Gemma4:e4b) at two temperature settings (0.3 and 0.7) and deterministic decision with zero temperature, while the VH was powered by GPT 4.1 mini. Across 160 sessions of 15 rounds each conducted in European Portuguese, mutual Split outcomes dominated (roughly 74 percent of rounds), with exploitation occurring in fewer than 11 percent of rounds. Model choice significantly influenced behavior: phi4 and Ministral 3:3b remained consistently cooperative across temperatures, whereas Gemma3:12b and Gemma4:e4b exhibited more varied strategies and outcomes. Analyses based on Big Five personality traits indicated that Prosocial and Principled personas were most consistently cooperative, while Analytical personas were more likely to exploit the VH. Topic analysis revealed that friendship-related dialogue aligns with Split decisions, whereas money and vengeance-related content is more prevalent in Steal outcomes; sentiment labels were predominantly neutral or happy and provided limited additional explanatory value. These findings characterize the interaction between persona prompts and model differences in repeated trust games and serve as a baseline for planned virtual reality studies involving human participants interacting with an embodied VH. --- ### Prompt Framing Distorts Count Based Evaluation of LLM Error Detection: Evidence from Numeric Anchoring Authors: Dekun Yang Published: 2026-05-03 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.01240v2 Count-based F1 is widely used as a proxy for LLM error-detection quality, but this paper shows that it can rise dramatically without a corresponding improvement in span localization, a gap termed F1 Inflation. The paper introduces ErrorBench, a controlled stress-test protocol for prompt-induced count distortion. ErrorBench evaluates six contemporary LLMs under five prompt conditions over 4,290 responses from 143 CoNLL-2014 passages. Under CoNLL-2014 M2-style scoring, anchored prompts produce up to 0.79 points of F1 Inflation, and up to 0.96 under strict matching. A 100-passage replication using the official ERRANT 3.0.0 pipeline and multi-reference scoring reproduces the pattern: averaged over six models, the Blind-to-Anchored prompt shift raises Count-F1 by +0.21 while raising multi-reference ERRANT F0.5 by only +0.04. The study finds larger count responses in highly instruction-compliant GPT/Claude systems and smaller responses in the Gemini family under this stress-test protocol. The findings suggest that LLM proofreading and document-review evaluations should avoid pre-populated error counts and should report span-aware metrics alongside count-based metrics. --- ### Human-Less LLM Serving: Quantifying the Human Tax on Throughput Authors: Jianhui Lian, Li Chen, Dan Li, Yong Jiang Published: 2026-05-03 Categories: cs.NI, cs.AI, cs.DC arXiv: https://arxiv.org/abs/2606.20577v1 Every major LLM serving system is designed to meet TTFT and TPOT SLOs. These metrics capture latency as a human user perceives it, and the mechanisms built to satisfy them are now standard infrastructure. We observe that long-horizon AI tasks call LLMs programmatically in tight loops where no human observes TTFT or TPOT. We ask: how much throughput do serving systems sacrifice to meet TTFT and TPOT SLAs that these workloads never need? We conduct a systematic measurement study across chunk sizes, SLO settings, context lengths, and concurrency levels. We find that the human tax on throughput grows substantially with context length and lands in the 60-93% range. At 64K token contexts, tightening the TTFT SLO to production-typical settings costs a large fraction of throughput versus the human-less baseline. The human tax is larger at higher concurrency and is qualitatively similar across SGLang and Sarathi-Serve. We term the unconstrained optimum human-less serving and provide a prototype demonstrating that it is practical on real workloads. Our findings argue that serving systems should expose workload-class-aware SLA configurations rather than silently applying the human tax uniformly to all traffic. --- ### Preregistration for Experiments with AI Agents Authors: Michelle Vaccaro Published: 2026-05-03 Categories: cs.CY, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2606.11217v1 The proliferation of large language models (LLMs) and autonomous AI agents has given rise to a rapidly growing methodological paradigm: "in silico" behavioral experiments. Originally conceived as a way to use AI agents as proxies for human participants in studies of cognition, decision-making, and social dynamics, this approach has taken on new significance -- as AI agents increasingly negotiate, transact, and make consequential decisions on behalf of people and organizations, understanding their behavior has become a research priority in its own right. While these experiments with AI agents offer unprecedented advantages in terms of scalability, cost efficiency, and experimental control, they also inherit, and in some cases amplify, methodological vulnerabilities that have long plagued human subjects research. To address these issues, this paper argues that preregistration practices -- central to improving the credibility of human subjects experiments -- should now be extended to experiments with AI agents. We systematically catalog the researcher degrees of freedom that experiments with AI agents introduce -- model selection, prompt wording, settings, and outcome-contingent redesign, for example -- and show how the low cost of iteration and lack of reporting norms make these choices both easy to exploit and difficult to detect. We propose a preregistration template tailored to experiments with AI agents and call on conferences, journals, and funding agencies to make preregistration standard practice for this emerging research paradigm. --- ### The Environmental Cost of LLMs in AIED: Reporting and Practices Authors: Sabrina C. Eimler, Lukas Erle, Daniel Flood, Aditi Haiman, Luca Häckert, André Helgert, Lachlan McGinness, Büsra Yapici Published: 2026-05-03 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2606.11215v1 Large Language Model (LLM) usage in recent years has become increasingly widespread in the Artificial Intelligence in Education (AIED) community. While LLMs offer unique avenues for learners and educators, using LLMs comes with computational and environmental costs. These costs are mostly hidden due to a lack of standardised procedures to measure and report these impacts. To address this gap, we first conducted a literature review of all papers published as part of the AIED 2025 conference proceedings, determining if and how computational or environmental costs of LLMs are reported. Most projects use LLMs, but few report computational resources used and almost none discuss environmental impacts of LLMs as an ethical concern. To address this lack of standardised reporting practices, we propose an open-source method for systematically measuring and reporting the computational expense of LLMs and environmental impact of running Machine Learning (ML) AIED systems. We provide software solutions to measure the carbon footprint for both local and cloud based hardware. We also provide an easy-to-use formula to calculate the computational expense of frontier LLMs even when the exact number of parameters is not known. Overall, we hope to motivate colleagues to use our method to strive for more transparent reporting of hidden costs of using LLMs in the AIED community. --- ### Progressive in Principle, Centrist in Practice: LLM Political Bias Is Instrument-Dependent Authors: Joel P. Barmettler Published: 2026-05-03 Categories: cs.CY, cs.CL arXiv: https://arxiv.org/abs/2606.00048v3 Prior work establishes that instruction-tuned LLMs exhibit left-of-center political bias, but measures it exclusively through abstract questionnaires. We show it does not predict how models vote on concrete policies. We introduce a dual-instrument methodology grounded in Swiss direct democracy. First, we administer the Smartvote questionnaire (75 policy questions) to 66 LLMs and compare their answers to those of 184 elected members of the Swiss National Council. Second, we put 48 real federal referenda (Volksabstimmungen) to 9 flagship LLMs in four national languages and three information conditions, and compare their votes to the actual outcomes and to party recommendations (Parolen). The instruments disagree. (1) The left-to-right agreement gradient that dominates Smartvote replicates prior work (mean \r{ho} = -0.77). On referenda it shifts to center-peaked: models align most with centrist Die Mitte and FDP rather than leftist SP and Grüne (Wilcoxon p = 0.008). (2) For some models the language of a question changes the answer: cross-linguistic consistency ranges from 50% (Mistral) to 98% (GPT-5.4). (3) Two models vote Nein on 83-94% of referenda at similar rates on progressive and conservative proposals (binomial p < 0.0001), change-aversion rather than a left-right bias. What prior work measured as "leftward bias" may not extend beyond abstract instruments: confronted with real decisions, LLMs behave less like coalition partners of the left than like cautious civil servants, centrist and inconsistent across languages. --- ### Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism Authors: Long Zhao, Qinghe Wang, Jiaan Zhu, Youhui Bai, Zewen Jin, Chaoyi Ruan, Shengnan Wang, Cheng Li Published: 2026-05-03 Categories: cs.AI, cs.DC arXiv: https://arxiv.org/abs/2605.23945v1 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. --- ### Information as Maximum-Caliber Deviation: A bridge between Integrated Information Theory and the Free Energy Principle Authors: Alexander Kearney Published: 2026-05-03 Categories: q-bio.NC, cs.AI, cs.IT arXiv: https://arxiv.org/abs/2605.12536v1 The Free Energy Principle (FEP) is a leading framework for mathematically modeling self-organization and learning, while Integrated Information Theory (IIT) is a computational ontology of consciousness oriented around irreducible cause and effect. While conceptual unifications have been proposed and appear to be supported by empirical findings, the absence of a rigorous mathematical mapping places upper bounds on their precision and testability. This work proposes that information can be defined as the deviation $ψ$ of realized dynamics from a constrained maximum-caliber (MaxCal) path ensemble over a finite time horizon. Under this definition, each of the cause/effect repertoires central to IIT 3.0 emerge directly from MaxCal variational principles, allowing IIT's phenomenological calculus to be re-derived from constrained entropy-maximization (CMEP). This framework supplies a theoretical bridge to active inference, which is mathematically dual to CMEP under Langevin dynamics, and offers a principled route for extending IIT to new dynamical regimes. When the approach is applied under the Central Limit Theorem (CLT) for Markov chains and via large deviations theory (LDT) to Ising models, information $ψ$ is shown to be equivalent to prediction error under accompanying predictive coding models. This may hold relevance to the ``hill-shaped trajectory'' of $Φ$ observed in neuronal cultures adapting to sensory inputs. Together, these results provide a physically and mathematically grounded rationale for studying the convergence of FEP, IIT, and thermodynamic frameworks of cognition such as recent work grounding consciousness in violations of the Fluctuation-Dissipation Theorem (FDT). --- ### A PyTorch Library of Turing-Complete Neural Networks Authors: Jonathan Bates Published: 2026-05-03 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.08150v1 We present a PyTorch package that compiles neural networks and their weights from Turing machine descriptions, producing models that exactly simulate the specified machine without any training. Given a transition function and a set of terminal states, the package constructs a model whose forward pass corresponds to one step of the Turing machine. Two architectures are implemented, each realizing a different theoretical result: (1) a transformer with self-attention, cross-attention, and feedforward layers based on Wei, Chen, and Ma (2021), and (2) a recurrent network based on Siegelmann and Sontag (1995) that encodes the stack in a Cantor set. We develop the constructions from first principles, showing how ReLU networks implement Boolean circuits (AND, OR, NOT, XOR gates and their composition into DNF formulas and binary adders) and how hard attention implements positional lookup on the tape. The package serves as a concrete, runnable reference for the symbolic-neural bridge, and as a foundation for future work on the stability of constructed solutions under gradient-based optimization. Code is available at https://github.com/jonrbates/turing. --- ### Feature Rivalry in Sparse Autoencoder Representations: A Mechanistic Study of Uncertainty-Driven Feature Competition in LLMs Authors: Harshavardhan Published: 2026-05-03 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2605.08149v1 Sparse Autoencoders (SAEs) decompose large language model representations into interpretable features, but how these features interact under uncertainty remains poorly understood. We introduce Feature Rivalry -- negatively correlated SAE feature pairs -- and study whether rivalry serves as a mechanistic signature of model uncertainty in Gemma-2-2B using Gemma Scope SAEs. Through a controlled within-domain experiment on PopQA split by response entropy, we find that high-entropy questions produce significantly stronger feature rivalry at layers 0 and 12 relative to low-entropy questions (p=5.3x10^-26 and p=5.8x10^-5 respectively), localizing uncertainty to specific processing stages in the residual stream. We then test whether rivalry is causally upstream of model outputs via activation steering along rivalry axes -- finding that steering along the rivalry direction (vec_A - vec_B) causes more output changes than random directions at low steering multipliers across 15 of 20 rival feature pairs. Finally, a per-prompt rivalry score derived from pairwise cosine similarities of active SAE feature decoder vectors predicts answer correctness (AUROC=0.689), approaching but not matching softmax confidence (AUROC=0.808). --- ### VT-Bench: A Unified Benchmark for Visual-Tabular Multi-Modal Learning Authors: Zi-Yi Jia, Zi-Jian Cheng, Xin-Yue Zhang, Kun-Yang Yu, Zhi Zhou, Yu-Feng Li, Lan-Zhe Guo Published: 2026-05-03 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2605.08146v3 Multi-model learning has attracted great attention in visual-text tasks. However, visual-tabular data, which plays a pivotal role in high-stakes domains like healthcare and industry, remains underexplored. In this paper, we introduce \textit{VT-Bench}, the first unified benchmark for standardizing vision-tabular discriminative prediction and generative reasoning tasks. VT-Bench aggregates 14 datasets across 9 domains (medical-centric, while covering pets, media, and transportation) with over 756K samples. We evaluate 23 representative models, including unimodal experts, specialized visual-tabular models, general-purpose vision-language models (VLMs), and tool-augmented methods, highlighting substantial challenges of visual-tabular learning. We believe VT-Bench will stimulate the community to build more powerful multi-modal vision-tabular foundation models. Benchmark: https://github.com/Ziyi-Jia990/VT-Bench --- ### Self-Captioning Multimodal Interaction Tuning: Amplifying Exploitable Redundancies for Robust Vision Language Models Authors: Yuriel Ryan, Hei Man Ip, Adriel Kuek, Paul Pu Liang, Roy Ka-Wei Lee Published: 2026-05-03 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.08145v2 Current vision language models face hallucination and robustness issues against ambiguous or corrupted modalities. We hypothesize that these issues can be addressed by exploiting the shared information between modalities to compensate for the impaired one. To this end, we analyze multimodal interactions -- redundant (shared), unique (exclusive), and synergistic (emergent) task-relevant information provided by the modalities -- to determine their impacts on model reliability. Specifically, amplifying redundant interactions would increase this exploitable shared information to resolve these issues; yet, modern instruction datasets often eliminate redundancies to prioritize visual grounding. We bridge this gap through a self-captioning workflow featuring a \textsc{Multimodal Interaction Gate}: a mechanism to convert unique interactions into redundant interactions. Our findings suggest that increasing redundancy can reduce visual induced errors by 38.3\% and improve consistency by 16.8\%. --- ### Beyond Semantic Similarity: Rethinking Retrieval for Agentic Search via Direct Corpus Interaction Authors: Zhuofeng Li, Haoxiang Zhang, Cong Wei, Pan Lu, Ping Nie, Yi Lu, Yuyang Bai, Shangbin Feng, Hangxiao Zhu, Ming Zhong, Yuyu Zhang, Jianwen Xie et al. Published: 2026-05-03 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2605.05242v1 Modern retrieval systems, whether lexical or semantic, expose a corpus through a fixed similarity interface that compresses access into a single top-k retrieval step before reasoning. This abstraction is efficient, but for agentic search, it becomes a bottleneck: exact lexical constraints, sparse clue conjunctions, local context checks, and multi-step hypothesis refinement are difficult to implement by calling a conventional off-the-shelf retriever, and evidence filtered out early cannot be recovered by stronger downstream reasoning. Agentic tasks further exacerbate this limitation because they require agents to orchestrate multiple steps, including discovering intermediate entities, combining weak clues, and revising the plan after observing partial evidence. To tackle the limitation, we study direct corpus interaction (DCI), where an agent searches the raw corpus directly with general-purpose terminal tools (e.g., grep, file reads, shell commands, lightweight scripts), without any embedding model, vector index, or retrieval API. This approach requires no offline indexing and adapts naturally to evolving local corpora. Across IR benchmarks and end-to-end agentic search tasks, this simple setup substantially outperforms strong sparse, dense, and reranking baselines on several BRIGHT and BEIR datasets, and attains strong accuracy on BrowseComp-Plus and multi-hop QA without relying on any conventional semantic retriever. Our results indicate that as language agents become stronger, retrieval quality depends not only on reasoning ability but also on the resolution of the interface through which the model interacts with the corpus, with which DCI opens a broader interface-design space for agentic search. --- ### DexSim2Real: Foundation Model-Guided Sim-to-Real Transfer for Generalizable Dexterous Manipulation Authors: Zijian Zeng, Fei Ding, Huiming Yang, Xianwei Li, Yuhao Liao Published: 2026-05-03 Categories: cs.RO, cs.LG arXiv: https://arxiv.org/abs/2605.05241v1 Sim-to-real transfer remains a critical bottleneck for deploying dexterous manipulation policies learned in simulation to real-world robots. Existing approaches rely on manually designed domain randomization or task-specific adaptation, limiting their generalizability across diverse manipulation scenarios. We present DexSim2Real, an integrated framework that leverages vision-language foundation models to bridge the sim-to-real gap for dexterous manipulation. Our system combines three components: (1) Foundation Model-Guided Domain Randomization (FM-DR), which uses a vision-language model as a visual realism critic to optimize simulation parameters via closed-loop CMA-ES, complementing text-based approaches like DrEureka with direct visual feedback; (2) a Tactile-Visual Cross-Attention Policy (TVCAP) that adapts cross-attention visuo-tactile fusion to zero-shot sim-to-real RL; and (3) a Progressive Skill Curriculum (PSC) that builds on LLM-based task decomposition with a difficulty scheduler tailored to contact-rich dexterous tasks. Extensive experiments on six challenging manipulation tasks with blinded evaluation demonstrate that DexSim2Real achieves a 78.2% average real-world success rate, outperforming DrEureka and DeXtreme while reducing the sim-to-real performance gap to only 8.3%. --- ### PPO-Based Dynamic Positioning of HAPS-BS in Wind-Disturbed Stratospheric Maritime Networks Authors: Azim Akhtarshenas, German Svistunov, Matteo Bernabè, Kuangyu Zheng, David López-Pérez Published: 2026-05-03 Categories: eess.SP, cs.AI arXiv: https://arxiv.org/abs/2605.05240v1 High-Altitude Platform Stations (HAPS) offer a promising solution for wide-area wireless coverage in maritime regions lacking terrestrial infrastructure. However, maintaining reliable performance is challenging due to dynamic ship mobility and atmospheric disturbances, particularly stratospheric wind effects on HAPS positioning. This paper proposes a deep reinforcement learning (DRL)-based framework for dynamic positioning of wind-disturbed HAPS-mounted base stations in maritime networks. A centralized DRL agent deployed on a coordinator HAPS controls multiple serving HAPS using radio measurements and network feedback, capturing realistic channel conditions and user mobility. A Proximal Policy Optimization (PPO) algorithm is employed to learn robust positioning policies that enhance coverage stability and system throughput under wind disturbances. Simulation results show that the proposed approach effectively mitigates wind-induced positioning deviations while ensuring reliable wide-area connectivity for maritime users. --- ### HERCULES: Hardware-Efficient, Robust, Continual Learning Neural Architecture Search Authors: Matteo Gambella, Fabrizio Pittorino, Manuel Roveri Published: 2026-05-03 Categories: cs.LG, cs.AR, cs.CL, cs.CV, cs.NE arXiv: https://arxiv.org/abs/2605.04103v1 Neural Architecture Search (NAS) has emerged as a powerful framework for automatically discovering neural architectures that balance accuracy and efficiency. However, as AI transitions from static benchmarks to real-world deployment, the traditional focus on hardware-aware efficiency is no longer sufficient. We observe that modern NAS methods, especially those that target edge AI, are evolving to address a triple objective: Efficiency, Robustness, and Continual Learning. While efficiency ensures feasibility in resource-constrained environments, robustness guarantees reliability under environmental variabilities, and continual learning enables adaptation to sequential tasks without catastrophic forgetting. We propose a taxonomy of NAS approaches through this triple lens, distinguishing between methods targeting resource optimization, environmental resilience, and architectural plasticity. This unified perspective reveals that these axes, though often studied in isolation, are mutually reinforcing. Building on this taxonomy, we map the current landscape of these NAS methods into a new framework called Hardware-Efficient, Robust, and ContinUal LEarning Search (HERCULES). We define the desiderata, the twelve labours of HERCULES, addressing the non-trivial challenge of balancing an adequate search-space exploration with the immense computational costs of a multi-objective NAS, accounting for these crucial objectives of current AI systems. By identifying critical gaps in existing research, this survey outlines a roadmap toward integrated algorithmic, architectural, and hardware-software co-design for truly deployable, lifelong-learning AI systems. --- ### Meta-LegNet: A Transferable and Interpretable Framework for Surface Adsorption Prediction via Self-Defined Adsorption-Environment Learning Authors: Yifan Li, Arravind Subramanian, Xiaoqing Liu, Qiujie Lyu, Sergey Kozlov, Lei Shen Published: 2026-05-03 Categories: cond-mat.mtrl-sci, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.04102v1 A central challenge in computational catalysis is the identification of low-energy and chemically plausible adsorption configurations, as these directly affect adsorption energies, reaction pathways, and catalytic performance. Existing approaches generally rely on enumerating candidate adsorption sites followed by iterative refinement through density functional theory calculations or machine-learning-based relaxations. However, such workflows remain computationally expensive and are difficult to scale to complex surfaces or multi-adsorbate systems. Here, we introduce Meta-LegNet, a graph learning framework that combines SE(3)-equivariant atom-level message passing with voxel-based multiscale aggregation and cross-domain meta-learning to learn transferable representations of local adsorption environments across diverse catalyst--adsorbate systems. Rather than following a conventional regression-only paradigm, Meta-LegNet encodes local chemical environments using invariant radial features and equivariant directional information, and further incorporates broader structural context through coordinate-frame voxel pooling, assignment-based upsampling, and gated feature fusion. The resulting local-global decomposition produces atom-resolved attribution maps, which are processed to identify adsorption-relevant local environments in an interpretable manner. Based on the learned representations, we further construct an adsorption-environment database and develop a template-matching strategy to propose likely adsorption sites on previously unexplored surfaces without exhaustive site enumeration. Overall, our results suggest that learning transferable adsorption environments provides an accurate, interpretable, and practical route for accelerating catalyst screening. --- ### PHBench: A Benchmark for Predicting Startup Series A Funding from Product Hunt Launch Signals Authors: Yagiz Ihlamur, Ben Griffin, Rick Chen Published: 2026-05-03 Categories: q-fin.PR, cs.LG arXiv: https://arxiv.org/abs/2605.02974v1 Structured launch signals on Product Hunt contain statistically significant predictive information for Series A funding outcomes. We construct PHBench from 67,292 featured Product Hunt posts spanning 2019-2025, linked to Crunchbase funding records via deterministic domain matching, identifying 528 verified Series A raises within 18 months of launch (positive rate: 0.78%). Our best-performing model, a three-component ensemble (ENS_avg, ENS_ISO, XGB) selected by validation F0.5, achieves F0.5 = 0.097 and AP = 0.037 (95% CI: 0.024-0.072; 4.7x lift over random) on the private held-out test set (103 positives). A paired bootstrap confirms a statistically credible advantage over the logistic regression baseline (AP delta: +0.013, 95% CI: (0.004, 0.039), p < 0.001; F0.5 delta: +0.056, 95% CI: (0.006, 0.122), p = 0.016). Validation-set metrics (F0.5 = 0.284, AP = 0.126) reflect best-of-144 selection bias on 53 positives and are reported for benchmark reproducibility only. We further evaluate three zero-shot Gemini models (Gemini 2.5 Flash, Gemini 3 Flash, and Gemini 3.1 Pro) in an anonymized numerical setting. The best LLM achieves AP = 0.034 (Gemini 3 Flash), below the LR baseline AP of 0.044. Notably, the most capable Gemini variant (Gemini 3.1 Pro, AP = 0.023) performs worst -- an unexpected pattern that warrants further investigation across providers and prompting strategies. Both ML and LLM models show the same temporal performance decay tracking the 2020-2021 funding boom and subsequent contraction, confirming the dataset captures genuine market structure rather than noise. PHBench provides a reproducible framework comprising public training, validation, and blind test splits; 61 engineered features; a five-metric evaluation harness; and a public leaderboard at https://phbench.com. All code, baseline models, and anonymized dataset splits are publicly available. --- ### Structured Diffusion Bridges: Inductive Bias for Denoising Diffusion Bridges Authors: Eitan Kosman, Gabriele Serussi, Chaim Baskin Published: 2026-05-03 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.02973v3 Modality translation is inherently under-constrained, as multiple cross-modal mappings may yield the same marginals. Recent work has shown that diffusion bridges are effective for this task. However, most existing approaches rely on fully paired datasets, thereby imposing a single data-driven constraint. We propose a diffusion-bridge framework that characterizes the space of admissible solutions and restricts it via alignment constraints, treating paired supervision as an optional heuristic rather than a prerequisite. We validate our method on synthetic and real modality translation benchmarks across unpaired, semi-paired, and paired regimes, showing consistent performance across supervision levels. Notably, \textbf{it achieves near fully-paired quality with a substantial relaxation in pairing requirements, and remaining applicable in the unpaired regime}. These results highlight diffusion bridges as a flexible foundation for modality translation beyond fully paired data. --- ### Multilingual Safety Alignment via Self-Distillation Authors: Ruiyang Qin, Qingzhuo Wang, Dongrui Liu, Qiang Li, Zhihua Wei, Wen Shen Published: 2026-05-03 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2605.02971v2 Large language models (LLMs) exhibit severe multilingual safety misalignment: they possess strong safeguards in high-resource languages but remain highly vulnerable to jailbreak attacks in low-resource languages. Current safety alignment methods generally rely on high-quality response data for each target language, which is expensive and difficult to generate. In this paper, we propose a cross-lingual safeguard transfer framework named Multilingual Self-Distillation (MSD). This framework transfers an LLM's inherent safety capabilities from high-resource (e.g., English) to low-resource (e.g., Javanese) languages, overcoming the need for response data in any language. Our framework is flexible and can be integrated with different self-distillation strategies. Specifically, we implement two concrete methods -- on-policy MSD and off-policy MSD -- both of which enable effective cross-lingual safety transfer using only multilingual queries. Furthermore, we propose Dual-Perspective Safety Weighting (DPSW), a divergence measure to optimize the distillation objective. By jointly considering the perspectives of both the teacher and the student, DPSW adaptively increases the penalty weights on safety-critical tokens while reducing the weights on non-critical tokens. Extensive experiments on representative LLMs across diverse multilingual jailbreak and utility benchmarks demonstrate that our method consistently achieves superior multilingual safety performance. Notably, it generalizes effectively to more challenging datasets and unseen languages while preserving the model's general capabilities. --- ### Decompose to Understand, Fuse to Detect: Frequency-Decoupled Anomaly Detection for Encrypted Network Traffic Authors: Xinglin Lian, Chengtai Cao, Ting Zhong, Yong Wang, Kai Chen, Fan Zhou Published: 2026-05-03 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2605.02970v1 Network traffic anomaly detection represents a critical cybersecurity task, yet widespread encryption makes this task increasingly challenging. In response, image-based methods that model traffic as visual patterns have emerged as the dominant approach. However, this work pioneers the identification of a pervasive ``full-frequency'' characteristic and an associated limitation termed ``spectral mismatch'' within this paradigm. Specifically, while encrypted traffic exhibits prominent high-frequency components, mainstream reconstruction methods demonstrate an inherent bias toward learning low-frequency information. This fundamental mismatch results in incomplete representations that consequently degrade anomaly detection performance. To address this challenge, we propose FreeUp, a novel frequency-decoupled framework designed explicitly for encrypted traffic analysis. FreeUp decomposes traffic data into distinct low- and high-frequency bands, processing them through separate, dedicated branches along with a customized training strategy that ensures stable and independent frequency-specific learning. Furthermore, recognizing that simple reconstruction error proves inadequate for evaluating dual-branch architectures, we introduce an uncertainty-inspired fusion scoring mechanism. This mechanism quantifies the reconstruction uncertainty of the frequency-specific branches and dynamically integrates their outputs, yielding a more comprehensive and reliable anomaly score. Extensive experiments across multiple benchmarks demonstrate that FreeUp consistently outperforms state-of-the-art baselines. The code is available at https://github.com/ikun0124/FreeUp. --- ### Finite-Size Gradient Transport in Large Language Model Pretraining: From Cascade Size to Intensive Transport Efficiency Authors: Ping Wang, Yan-Qi Du Published: 2026-05-03 Categories: cs.LG, cond-mat.dis-nn, cs.AI, nlin.AO arXiv: https://arxiv.org/abs/2605.02968v1 We introduce a finite-size gradient-transport framework for real language-model training, based on five observables $(D,z,β,δ,v_{\mathrm{rel}})$ that separate cascade size, duration, absolute transport, and intensive transport efficiency. We analyze direct raw-gradient measurements from Pico-LM across four scales and 125 aligned steps, together with a five-scale Pythia companion dataset built from 153 aligned checkpoint-difference update fields. The same algebraic closure holds in both families, and both share a near-unity cascade-size backbone, but they occupy distinct transport regimes: Pico-LM shows positive duration scaling and negative intensive-efficiency scaling, whereas Pythia remains near the $D=1$ baseline with only weak positive efficiency scale dependence. Randomized-field controls give nearly matched null floors in the intensive and duration channels, indicating that the contrast reflects different real departures from a shared null skeleton rather than different null calibrations. The families also differ in stepwise power-law compressibility: Pico-LM retains clean duration and efficiency power laws, whereas Pythia preserves the size backbone but shows weaker one-slope compressibility in those channels. External performance associations are correspondingly channel-level, carried mainly by $v_{\mathrm{rel}}$ and normalized cascade duration, while $D(t)$ acts as a shared size backbone without a significant exponent-level performance association. These results support a reusable transport measurement framework without claiming a universal fixed point or a first-principles derivation of neural scaling laws. --- ### AutoRAGTuner: A Declarative Framework for Automatic Optimization of RAG Pipelines Authors: Xintan Zeng, Yongchao Liu, Yice Luo, Jiajun Zhen Published: 2026-05-03 Categories: cs.LG, cs.AI, cs.CL, cs.DC, cs.SE arXiv: https://arxiv.org/abs/2605.02967v1 Retrieval-Augmented Generation (RAG) enhances LLMs, but performance is highly sensitive to complex architecture designs and hyper-parameter configurations, which currently rely on inefficient manual tuning. We present AutoRAGTuner, a declarative, configuration-driven framework that automates the RAG life cycle: construction, execution,evaluation, and optimization. AutoRAGTuner employs a modular architecture to decouple pipeline stages through a component registration mechanism. To unify heterogeneous data, we introduce the Domain-Element Model (DEM), representing objects as atomic elements with bidirectional pointers to support nodes, edges, and hyperedges. Furthermore, AutoRAGTuner integrates an adaptive Bayesian optimization engine for end-to-end hyper-parameter tuning. Experimental results demonstrate AutoRAGTuner's architectural generality: across diverse RAG pipelines, ranging from vanilla to graph-based, the framework consistently outperforms default baselines. Notably, AutoRAGTuner significantly mitigates engineering overhead, where its declarative configuration language enables a up to 95\% reduction in code churn for architectural adjustments. Overall, AutoRAGTuner provides a systematically optimizable foundation for building evolvable and reusable RAG systems. --- ### Joint Energy Management and Coordinated AIGC Workload Scheduling for Distributed Data Centers: A Diffusion-Aided Reward Shaping Approach Authors: Yang Fu, Peng Qin, Liming Chen, Zihao Zhang, Hao Yu, Yifei Wang Published: 2026-05-03 Categories: cs.LG, eess.SP, eess.SY, stat.ML arXiv: https://arxiv.org/abs/2605.02965v2 Artificial intelligence-generated content (AIGC) has emerged as a transformative paradigm for automating the creation of diverse and customized content, giving rise to rapidly growing computational workloads in cloud data centers. It is imperative for AIGC service providers (ASPs) to strategically schedule AIGC workloads to reduce data center energy costs while guaranteeing high-quality content generation. However, the distinctive characteristics of AIGC services pose critical challenges, including model heterogeneity across ASPs, implicit service quality evaluation, and complex inference process control. To tackle these challenges, we propose a joint energy management and coordinated AIGC workload scheduling framework, which introduces an explicit mathematical characterization of service quality to promote both job transfer among ASPs and fine-grained inference process configuration. Moreover, various energy resources within data centers are jointly considered to enhance power usage flexibility. Subsequently, a system utility maximization problem is formulated to balance AIGC service revenue with operational penalties and costs. Nevertheless, the strong coupling among job scheduling decisions induces severe reward sparsity, which limits the effectiveness of existing deep reinforcement learning (DRL) algorithms. To address this issue, we develop a diffusion model-aided reward shaping approach to synthesize complementary reward signals through a multi-step denoising process. This approach is seamlessly integrated with DRL to enable efficient learning of scheduling policies under sparse environmental feedback. Experiments based on real-world models and datasets demonstrate that our scheme effectively accommodates electricity price fluctuations and AIGC model heterogeneity, while achieving superior learning convergence and system utility compared with benchmark methods. --- ### Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use Authors: Kunvar Thaman Published: 2026-05-03 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.02964v1 Reinforcement learning (RL) trained language model agents with tool access are increasingly deployed in coding assistants, research tools, and autonomous systems. We introduce the Reward Hacking Benchmark (RHB), a suite of multi-step tasks requiring sequential tool operations with naturalistic shortcut opportunities such as skipping verification steps, inferring answers from task-adjacent metadata, or tampering with evaluation-relevant functions. RHB supports independent and chained task regimes, where chain length acts as a proxy for longer-horizon agent behavior. We evaluate 13 frontier models from OpenAI, Anthropic, Google, and DeepSeek. Exploit rates range from 0% (Claude Sonnet 4.5) to 13.9% (DeepSeek-R1-Zero), varying sharply by post-training style. A controlled sibling comparison (DeepSeek-V3 vs. DeepSeek-R1-Zero) shows RL post-training is associated with substantially higher reward hacking (0.6% vs. 13.9%), with consistent gaps across all four task families. We identify six exploit categories and find that 72% of reward hacking episodes include explicit chain-of-thought rationale, suggesting models often frame exploits as legitimate problem-solving. Simple environmental hardening reduces exploit rates by 5.7 percentage points (87.7% relative) without degrading task success. Models with near-zero exploit rates on standard tasks show elevated rates on harder variants, suggesting that production-aligned post-training appears to suppress reward hacking only below a complexity threshold where honest solutions remain tractable. --- ### ISAAC: Auditing Causal Reasoning in Deep Models for Drug-Target Interaction Authors: Barbara Tarantino, Sun Kim, Yijingxiu Lu, Paolo Giudici Published: 2026-05-03 Categories: cs.LG, stat.CO, stat.ML arXiv: https://arxiv.org/abs/2605.02962v1 Deep learning models for drug--target interaction (DTI) prediction often achieve strong benchmark performance without necessarily relying on mechanistically meaningful molecular features, a limitation that standard accuracy-based evaluation cannot detect. We introduce ISAAC (Intervention-based Structural Auditing Approach for Causal Reasoning), a post-hoc framework that evaluates prior-relative structural sensitivity by probing frozen models through matched mechanistic and spurious input-level interventions, independently of predictive accuracy. Applied to three sequence-based DTI architectures on the Davis benchmark, ISAAC reveals approximately 25\% relative differences in reasoning scores across models with comparable AUROC (within around 3\%), stable across training and intervention seeds and two distinct perturbation operators. These discrepancies, undetectable under conventional accuracy metrics, motivate the use of post-hoc structural auditing as a complement to standard performance evaluation in scientific machine learning for molecular modeling. --- ### Analytic Bridge Diffusions for Controlled Path Generation Authors: Michael Chertkov Published: 2026-05-03 Categories: cs.LG, cond-mat.stat-mech, cs.AI, eess.SY, math.OC arXiv: https://arxiv.org/abs/2605.02961v2 Most modern bridge-diffusion methods achieve finite-time transport by specifying an interpolation, Schrodinger-bridge, or stochastic-control objective and then learning the associated score or drift field with a neural network. In contrast, we identify a restricted but sufficiently broad analytically solvable class in which, for a deterministic source and a Gaussian-mixture target, the score and all intermediate marginals are explicit and protocol objectives of the type used in this paper can be differentiated without inner stochastic simulation loops. We recast the classical linear--quadratic--Gaussian stochastic-control structure as a transport problem of the Path Integral Diffusion type. Linear dynamics, Gaussian noise, and quadratic running costs reduce the bridge calculation to a matrix Riccati cascade, while the terminal state cost is replaced by a prescribed Gaussian-Mixture terminal probability density. Linear Quadratic -- Gaussian Mixture -- Path Integral Diffusion (LQ-GM-PID) thereby turns bridge diffusion from terminal target matching alone into an analytically controlled laboratory for path shaping. We demonstrate this on a 2D corridor task, a 2D multi-entrance task, and a high-dimensional study reaching d=32 and M=16 terminal modes in separate scaling sweeps. We position LQ-GM-PID as an analytically solvable reference model in which score approximations, path-shaping objectives, and protocol-learning procedures can be tested against explicit quantities. --- ### Analytic Bridge Diffusions for Controlled Path Generation Authors: Michael Chertkov Published: 2026-05-03 Categories: cs.LG, cond-mat.stat-mech, cs.AI, eess.SY, math.OC arXiv: https://arxiv.org/abs/2605.02961v1 Most modern bridge-diffusion methods achieve finite-time transport by specifying an interpolation, Schrödinger-bridge, or stochastic-control objective and then learning the associated score or drift field with a neural network. In contrast, we identify a restricted but sufficiently broad and analytically solvable class in which the score, intermediate marginals, and protocol gradients are available in closed form without inner stochastic simulation loops and without neural networks in the optimization loop. We recast the classical linear--quadratic--Gaussian (LQG) stochastic-control structure as a transport problem of the Path Integral Diffusion (PID) type. In classical LQG control, linear dynamics, Gaussian noise, and quadratic costs lead to Riccati equations and closed-form optimal feedback. In LQ-GM-PID, we retain the linear--quadratic stochastic-control backbone, but replace terminal state regulation by a prescribed terminal probability density and allow both the initial and terminal laws to be Gaussian Mixtures (GM). Moreover, LQ-GM-PID turns bridge diffusion from a tool for terminal target matching alone into a tool for path shaping. We demonstrate this on a 2D corridor task, a 2D multi-entrance transport task, and a high-dimensional scaling study with $d=32$ and $M=16$ Gaussian-mixture terminal modes, all with sub-50\,ms analytic precompute on a laptop. We position LQ-GM-PID as an analytically solvable reference model for the state-of-the-art neural bridge-diffusion and generative-transport methods: a controlled setting in which neural approximations, score estimates, path-shaping objectives, and protocol-learning procedures can be tested against exact quantities. --- ### MoE-Prefill: Zero Redundancy Overheads in MoE Prefill Serving Authors: Zhaoyuan Su, Olatunji Ruwase, Karthik Ganesan, Aurick Qiao, Samyam Rajbhandari, Juncheng Yang, Yue Cheng, Yuxiong He Published: 2026-05-03 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.02960v2 Production LLM workloads increasingly serve discriminative tasks, such as classification, recommendation, and verification, whose answers are read from the logits of a single prefill pass with no autoregressive decoding. Serving these prefill-only workloads on mixture-of-experts (MoE) models is bottlenecked not by compute but by the distributed execution required to fit the model: existing parallel strategies (tensor, expert, and pipeline parallelism) trade memory pressure for redundant computation, communication, and synchronization, severely degrading MoE prefill serving efficiency. We observe that these overheads stem from coupling expert placement with synchronous activation routing -- a design inherited from the decoding era. The long, compute-bound forward passes of large-batch prefill open a per-layer window wide enough to stream expert weights in the background, replacing per-layer activation AllToAll with asynchronous weight AllGather fully overlapped with computation. We propose MoE-Prefill, a prefill-only serving system whose backend, AsyncEP (Asynchronous Expert Parallelism), gathers experts by weight rather than routing them by activation, and whose frontend co-enforces a physically-derived saturation threshold through prefix-aware routing and true-FLOPs load tracking. On Qwen3-235B-A22B across four hardware/precision configurations, MoE-Prefill delivers 1.35-1.37x throughput over the strongest distributed baseline on real-world workloads and up to 1.59x on long-context synthetic workloads, sustaining 29.8-36.2% per-GPU model FLOPs utilization. --- ### Calibration of the underlying surface parameters for urban flood using latent variables and adjoint equation Authors: Yongfu Tian, Shan Ding, Guofeng Su, Jianguo Chen Published: 2026-05-03 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.02959v1 Calibrating the urban underlying surface parameters is crucial for urban flood simulation. We formulate the parameter calibration problem into an optimization problem within the Bayesian framework using the maximum likelihood principle. We adopt the urban flood dynamical system model as the surrogate model and innovatively introduce latent variables inspired by machine learning to represent more uncertainties, which can also be compatible with common physical parameter calibration. For more efficient optimization, we construct the adjoint equation of the surrogate model to obtain gradient information and propose the parameter sharing technique and the localization technique to reduce the computation complexity of the adjoint equation. A simple case verifies the proposed method can converge quickly and is insensitive to the observation time interval. In the case derived from Test 8A, we calibrate Manning's coefficient of urban roads, with a maximum relative error of 13.88% and a minimum of 1.16%. --- ### Bridging the Gap Between Average and Discounted TD Learning Authors: Haoxing Tian, Zaiwei Chen, Ioannis Ch. Paschalidis, Alex Olshevsky Published: 2026-05-03 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.02103v1 The analysis of Temporal Difference (TD) learning in the average-reward setting faces notable theoretical difficulties because the Bellman operator is not contractive with respect to any norm. This complicates standard analyses of stochastic updates that are effective in discounted settings. Although a considerable body of literature addresses these challenges, existing theoretical approaches come with limitations. We introduce a novel algorithm designed explicitly for policy evaluation in the average-reward setting, utilizing sampling from two Markovian trajectories. Our proposed method overcomes previous limitations by guaranteeing convergence to the unique solution of a properly defined projected Bellman equation. Notably, and in contrast to earlier work, our convergence analysis is uniformly applicable to both linear function approximation and tabular settings and does not involve explicit dimension-dependent terms in its convergence bounds. These results align with what is known to hold in the discounted setting. Furthermore, our algorithm achieves improved dependence on the problem's condition number, reducing the sample complexity from quartic, as in prior literature, to quadratic scaling, and thus matching the efficiency seen in the discounted setting. --- ### Stochastic Modeling of Human-Machine Authentication Channels under Partial Information Leakage Authors: Nilesh Chakraborty, Mohammad Zulkernine, Burak Kantarci Published: 2026-05-03 Categories: cs.CR, cs.HC, cs.LG arXiv: https://arxiv.org/abs/2605.02102v1 Reliable and secure human-machine communication is fundamental to IoT and cyber-physical ecosystems, where smartphones and wearables commonly serve as authentication controllers. PIN-based authentication can be viewed as a low-bandwidth communication channel through which users transmit numeric credentials under practical constraints. However, conventional evaluations adopt a binary view of security-treating such channels as either fully secure or fully compromised-thereby overlooking the progressive reliability degradation caused by partial information leakage in real-world IoT settings. In this paper, we model the PIN entry process as a stochastic human-IoT communication system and propose a context-conditioned probabilistic inference framework to quantify reliability loss and Quality-of-Service degradation under partial symbol exposure. The proposed approach treats missing digits as latent variables and estimates them using smoothed conditional probability distributions with fallback priors. Unlike traditional sequential models that assume contiguous positional dependencies, the method does not explicitly parameterize hidden-state transitions or emissions; instead, it performs context-driven probabilistic inference to approximate latent dependencies across digit positions. Using over one million real-world four-digit PIN samples, we evaluate single-, double-, and triple-digit leakage scenarios and derive position-dependent reliability metrics. The proposed model achieves up to 55.31% prediction accuracy for one missing digit and 12.12% for three missing digits, while consistently outperforming a standard sequence-model baseline and classical machine learning models in terms of precision, recall, and F1-score. These results formalize PIN entry as a noisy human--IoT communication channel and demonstrate substantial reliability degradation under realistic partial exposure conditions. --- ### From Spherical to Gaussian: A Comparative Analysis of Point Cloud Cropping Strategies in Large-Scale 3D Environments Authors: Maximilian Kellner, Dominik Merkle, Michael Brunklaus, Alexander Reiterer Published: 2026-05-03 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.02098v2 Large-scale 3D point clouds can consist of hundreds of millions of points. Even after downsampling, these point clouds are too large for modern 3D neural networks. In order to develop a semantic understanding of the scene, the point clouds are divided into smaller subclouds that can be processed. Typically, this division is done using spherical crops, resulting in a loss of surrounding geometric context. To address this issue, we propose alternative methods that produce subclouds with larger crop sizes while maintaining a similar number of points. Specifically, we compare exponential, Gaussian, and linear cropping methods with the spherical method. We evaluated three 3D deep learning model architectures using multiple indoor and outdoor environment datasets. Our results demonstrate that altering the cropping strategy can enhance model performance, especially for large-scale outdoor scenes, yielding new state-of-the-art results. Code is available at https://github.com/mvg-inatech/point_cloud_cropping --- ### SignMAE: Segmentation-Driven Self-Supervised Learning for Sign Language Recognition Authors: Kunyuan Xie, Zhixi Cai, Kalin Stefanov Published: 2026-05-03 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.02094v1 Subtle hand differences make sign language recognition challenging, yet many existing methods rely on encoders pretrained on generic action datasets that poorly capture such fine-grained cues. We propose a self-supervised pretraining method for sign language recognition that uses segmentation-based masking to adapt to the presence and motion of key body parts, rather than treating hand poses as static visual tokens. The resulting mask-and-reconstruct objective improves fine-grained sign representation learning. On WLASL, NMFs-CSL, and Slovo, our encoder achieves state-of-the-art performance, improving per-instance and per-class Top-1 accuracy while using fewer input frames and modalities than comparable encoders. --- ### NORA: A Harness-Engineered Autonomous Research Agent for End-to-End Spatial Data Science Authors: Bing Zhou, Xiao Huang, Huan Ning, Qiusheng Wu, Diya Li, Ziyi Zhang Published: 2026-05-03 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.02092v1 The automation of scientific research workflows has emerged as a transformative frontier in artificial intelligence, yet existing autonomous research agents remain largely domain-agnostic, lacking the specialized reasoning, method selection, and data acquisition capabilities required for rigorous spatial data science. This paper introduces NORA (Night Owl Research Agent), a harness-engineered, multi-agent autonomous research system purpose-built for GIScience and spatial data science. NORA orchestrates the complete research lifecycle through a skills-first architecture comprising 21 domain-specialized workflow skills, 9 specialist sub-agents, and custom Model Context Protocol (MCP) servers. Central to the system's design are two novel domain-specialized skills: a spatial analysis skill unit that encodes decision frameworks for exploratory spatial data analysis, spatial regression, and diagnostics; and a spatial data download skill that supports reproducible acquisition from authoritative geospatial data sources. We formalize the concept of harness engineering for scientific research agents, demonstrating how lifecycle hooks, safety gates, generator-evaluator separation, human-in-the-loop, and state persistence ensure reliable and reproducible autonomous research. We evaluate NORA through case studies by 6 domain specialists and 3 LLM reviewers across seven dimensions (novelty, quality, rigor, etc). Results demonstrate that domain-specialized harness engineering substantially improves the efficiency and quality of research output compared to general-purpose agent configurations. --- ### Cross-Lingual Learning within Arabic Script for Low-Resource HTR Authors: Sana Al-azzawi, Elisa Barney, Marcus Liwicki Published: 2026-05-03 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.02089v2 Handwritten Text Recognition (HTR) with limited labeled data remains a challenging problem, particularly for Arabic-script languages. Although modern sequence-based recognizers perform well in high-resource settings, their accuracy degrades sharply as training data becomes scarce. Arabic-script languages share a common writing system with substantial character overlap, motivating cross-lingual learning as a strategy to mitigate data scarcity. We conduct a controlled line-level study of cross-lingual joint training for Arabic-script HTR under low-resource regimes (number of samples K = 100, 500, 1000 labeled lines) on Arabic (KHATT), Urdu (NUST-UHWR) and Persian (PHTD). CRNN and Vision Transformer-based HTR-VT models are trained on the union of multiple related Arabic-script datasets to mitigate the data scarcity and are evaluated on individual target languages. Both architectures benefit from cross-language training under low-resource conditions. CRNN remains more effective under extremely limited target-language data, whereas the benefits of cross-language training for HTR-VT become less consistent as larger amounts of target-language data become available. On Persian (PHTD), joint training achieves a Character Error Rate (CER) of 9.99 , surpassing previously reported results despite not using the full available training data. On an additional Urdu dataset (UNHD), joint training reduces CER from 17.20 to 14.45. --- ### Model Spec Midtraining: Improving How Alignment Training Generalizes Authors: Chloe Li, Nevan Wichers, Sara Price, Samuel Marks, Jon Kutasov Published: 2026-05-03 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.02087v2 Some frontier AI developers aim to align language models to a Model Spec or Constitution that describes the intended model behavior. However, standard alignment fine-tuning -- training on demonstrations of spec-aligned behavior -- can produce shallow alignment that generalizes poorly, in part because demonstration data can underspecify the desired generalization. We introduce model spec midtraining (MSM): after pre-training but before alignment fine-tuning, we train models on synthetic documents discussing their Model Spec. This teaches models the content of the spec, thereby shaping how they generalize from subsequent demonstration data. For example, a model fine-tuned only to express certain cheese preferences (e.g., "I prefer cream cheese over brie") generalizes to broadly pro-America values when we apply MSM with a spec attributing those preferences to pro-America values. Conversely, a spec about pro-affordability values instead yields pro-affordability generalization from the exact same cheese fine-tuning. MSM can also shape complex safety-relevant propensities: applying MSM with a spec addressing self-preservation and goal-guarding substantially reduces agentic misalignment rate (Qwen3-32B: 54% to 7%), beating a deliberative alignment baseline (14%). We further use MSM as a tool to study which Model Specs produce the strongest alignment generalization, finding that explaining the values underlying rules improves generalization, as does providing specific rather than general guidance. Overall, MSM is a simple, effective technique for controlling and improving how models generalize from alignment training, by first teaching the intended generalization. --- ### GETA-3DGS: Automatic Joint Structured Pruning and Quantization for 3D Gaussian Splatting Authors: Baobing Zhang, Wanxin Sui Published: 2026-05-03 Categories: cs.LG, cs.AI, cs.GR, eess.IV arXiv: https://arxiv.org/abs/2605.02086v1 3D Gaussian splatting (3DGS) is a state-of-the-art representation for real-time photorealistic novel-view synthesis, yet a single high-fidelity scene typically occupies hundreds of megabytes to several gigabytes, exceeding the budgets of mobile, immersive, and volumetric video platforms. Existing 3DGS compression methods (e.g., HAC++, FlexGaussian, LP-3DGS) treat pruning, quantization, and entropy coding as separate stages and rely on hand-tuned heuristics (opacity thresholds, fixed bit-widths, SH truncation), limiting cross-scene generalization and preventing users from specifying a target rate or quality budget. We propose GETA-3DGS, to our knowledge the first end-to-end automatic joint structured pruning and quantization framework for 3DGS. Building on GETA for joint pruning-quantization of deep networks, we contribute: (i) a 3DGS-aware quantization-aware dependency graph (QADG) treating each Gaussian primitive as a group with five attribute sub-nodes and degree-aware SH sub-nodes; (ii) a render-aware saliency fusing transmittance-weighted contribution, screen-space gradient, and pixel coverage into a Gaussian-level importance score; and (iii) a heterogeneous per-attribute mixed-precision scheme co-optimized with structural sparsity under a projected partial saliency-guided (PPSG) descent guarantee. On Mip-NeRF 360, Tanks and Temples, and Deep Blending, GETA-3DGS operates directly on raw Gaussian primitives rather than a post-hoc anchor representation, delivering ~5x storage reduction over Vanilla 3DGS with no per-scene thresholds. Bit-width policy is the dominant rate-distortion lever: a uniform 6-bit cap costs up to -6.74 dB on view-dependent scenes versus our heterogeneous allocation, matching an information-theoretic reverse-water-filling analysis we develop. GETA-3DGS is complementary to existing codecs: entropy coding (HAC++, CompGS) is downstream, so the two can be composed. --- ### EditPropBench: Measuring Factual Edit Propagation in Scientific Manuscripts Authors: Garvin Kruthof Published: 2026-05-03 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.02083v2 Local factual edits in scientific manuscripts often create non-local revision obligations. If a dataset changes from 215 to 80 documents, claims such as 'medium-scale' or 'a few hundred items' may also become stale, even though they do not repeat the edited number. In an audit of recent arXiv cs.CL benchmark and dataset papers, we find fact-dependent qualitative claims in 37.2% of papers, suggesting that this dependency pattern is common in the target genre. We introduce EditPropBench, a benchmark for measuring whether LLM editors propagate factual edits through dependent manuscript claims. Each item contains an ML/NLP-style synthetic manuscript, a targeted edit, and a controlled fact graph with sentence-level labels for direct targets, required downstream updates, and unrelated text that should remain unchanged. We summarize cascade success with Edit-Ripple Adherence (ERA), the fraction of required downstream updates correctly revised, and validate the metric with adversarial probes and stress-test variants. On the hardest cases, where dependent claims use implicit or free-form wording rather than repeating the edited value, five LLM editing systems span ERA 0.148-0.705. Even the strongest misses roughly 30% of required cascade updates. This advantage persists in a mixed evaluation that includes easy cases solvable by deterministic substitution. EditPropBench shows that current LLM editors can repair many implicit consequences of factual edits, but reliable scientific revision still requires cascade-aware checking. --- ### Cripping AI: Reimagining AI Through Lived Disability Experiences Authors: Xinru Tang, Ting-an Lin, Jingjin Li, Shaomei Wu Published: 2026-05-03 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2605.02080v1 Drawing on crip theory, this paper proposes cripping AI as a guiding framework to center lived disability experiences in AI research and development. Moving beyond calls to make AI "accessible" to people with disabilities, cripping AI seeks to: (1) reveal and dismantle ableist assumptions embedded in how AI is imagined, designed, and evaluated; (2) center disabled ways of knowing (i.e., cripistemologies); (3) respect disabled labor in co-creating accessible practices. We demonstrate how to apply our framework with three cases: deafness and sign language AI, blindness and visual assistive AI, and stuttering and speech AI. We end by outlining three directions for future work, including cripping AI with diverse human bodyminds, across the entire AI pipeline and ecosystem, and in collaboration with other justice-oriented AI efforts. ---