# Research Papers — 2026-05 · bundle page 314 of 326 > Full-text markdown of papers #15651–15700 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/314 [← Previous bundle](https://aipod.dev/papers/archive/2026-05/page/313.md) [Next bundle →](https://aipod.dev/papers/archive/2026-05/page/315.md) ### Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels Authors: Plawan Kumar Rath, Rahul Maliakkal Published: 2026-05-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.15208v1 Large Language Models are routinely compressed via post-training quantization to reduce inference costs and memory footprint for cloud and edge deployment, yet the impact of this compression on model quality remains poorly understood. Existing studies typically compare only two conditions (full-precision vs. a single quantized variant), rely on aggregate bias metrics, and evaluate a single model family, making it impossible to distinguish gradual degradation from threshold-dependent safety failures. We conduct a controlled empirical study of three instruction-tuned models (Qwen2.5-7B, Mistral-7B, Phi-3.5-mini) at five precision levels (BF16 through 3-bit) on 12,148 BBQ bias benchmark items across 5 random seeds, totaling 911,100 inference records. Our results reveal that 3-bit quantization causes 6-21% of previously unbiased items to develop new stereotypical behaviors, following a clear dose-response pattern confirmed via logistic regression, while models' willingness to select "unknown" answers declines by 17.4%. Crucially, these item-level changes are invisible to standard quality metrics: perplexity increases by less than 0.5% at 8-bit and under 3% at 4-bit across all three models, yet 2.5-5.6% of items already develop new biases at 4-bit. These findings demonstrate that aggregate evaluation metrics systematically miss fairness-critical degradation, underscoring the need for quality-aware compression protocols that explicitly test for bias emergence before deployment. --- ### BioSEN: A Bio-acoustic Signal Enhancement Network for Animal Vocalizations Authors: Tianyu Song, Ton Viet Ta, Ngamta Thamwattana, Hisako Nomura, Linh Thi Hoai Nguyen Published: 2026-05-02 Categories: cs.SD, cs.LG, q-bio.NC arXiv: https://arxiv.org/abs/2605.12534v3 Most work in audio enhancement targets human speech, while bioacoustics is less studied due to noisy recordings and the distinct traits of animal sounds. To fill this gap, we adapt speech enhancement methods and build BioSEN, a model made for bioacoustic signals. BioSEN has three modules: a multi-scale dual-axis attention unit for time-frequency feature extraction, a bio-harmonic multi-scale enhancement unit for capturing harmonic structures, and an energy-adaptive gating connection unit that uses frequency weights to keep vocalizations from being removed as noise. Tests on three bioacoustic datasets show that BioSEN matches or exceeds state-of-the-art speech enhancement models while using far less computation. These results show BioSEN's strength for bioacoustic audio enhancement and its promise for biodiversity monitoring and conservation. --- ### NoiseRater: Meta-Learned Noise Valuation for Diffusion Model Training Authors: Fang Wu, Haokai Zhao, Da Xing, Hanqun Cao, Tinson Xu, Yanchao Li, Xiangru Tang, Zehong Wang, Aaron Tu, Kuan Pang, Hanchen Wang, Hongbin Lin et al. Published: 2026-05-02 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2605.08144v1 Diffusion models have achieved remarkable success across a wide range of generative tasks, yet their training paradigm largely treats injected noise as uniformly informative. In this work, we challenge this assumption and introduce NoiseRater, a meta-learning framework for instance-level noise valuation in diffusion model training. We propose a parametric noise rater that assigns importance scores to individual noise realizations conditioned on data and timestep, enabling adaptive reweighting of the training objective. The rater is trained via bilevel optimization to improve downstream validation performance after inner-loop diffusion updates. To enable efficient deployment, we further design a decoupled two-stage pipeline that transitions from soft weighting during meta-training to hard noise selection during standard training. Extensive experiments on FFHQ and ImageNet demonstrate that not all noise samples contribute equally, and that prioritizing informative noise improves both training efficiency and generation quality. Our results establish noise valuation as a complementary and previously underexplored axis for improving diffusion model training. Our code is available at: https://anonymous.4open.science/r/NoiseRater-DEB116. --- ### HoReN: Normalized Hopfield Retrieval for Large-Scale Sequential Model Editing Authors: Yuan Fang, Yi Xie, Xuming Ran Published: 2026-05-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.08143v2 Large language models encode vast factual knowledge that can become outdated or incorrect after deployment, yet retraining is prohibitively costly. This motivates lifelong model editing, which updates targeted behavior while preserving the rest of the model. Existing editors, both parameter-modifying and parameter-preserving, degrade severely as edits accumulate and struggle to generalize across paraphrases. We propose HoReN, a codebook-based parameter-preserving editor that wraps a single MLP layer with a discrete key-value memory. HoReN treats each codebook entry as both a knowledge key and a Hopfield stored pattern, retrieves edits by angular similarity on the unit hypersphere, and refines queries through damped Hopfield dynamics so paraphrases converge to the correct memory basin while unrelated inputs remain stable. HoReN achieves strong editing performance with consistent gains across diverse benchmarks spanning standard ZsRE, structured WikiBigEdit, and unstructured UnKE evaluations. Moreover, HoReN scales to 50K sequential edits on ZsRE with stable overall performance above 0.93, while prior editors collapse or degrade severely before reaching 10K. Our code is available at https://github.com/ha11ucin8/HoReN. --- ### Reasoning emerges from constrained inference manifolds in large language models Authors: Yanbiao Ma, Fei Luo, Linfeng Zhang, Chuangxin Zhao, Mingxuan Wang, Yinan Wu, Zhe Qian, Yang Lu, Long Chen, Zhao Cao, Xiaoshuai Hao, Ji-Rong Wen et al. Published: 2026-05-02 Categories: cs.LG, cs.CL, cs.CV arXiv: https://arxiv.org/abs/2605.08142v1 Reasoning in large language models is predominantly evaluated through labeled benchmarks, conflating task performance with the quality of internal inference. Here we study reasoning as an intrinsic dynamical process by examining the evolution of internal representations during inference. We find that inference-time dynamics consistently self-organize into low-dimensional manifolds embedded within high-dimensional representation spaces. we find that such geometric compression, although pervasive, is not sufficient for stable or reliable reasoning. Instead, effective reasoning dynamics emerge within a constrained structural regime characterized by three conditions: adequate representational expressivity, spontaneous manifold compression, and preservation of non-degenerate information volume within the compressed subspace. Models outside this regime exhibit characteristic pathological inference dynamics. Based on these insights, we introduce a unified, label-free diagnostic computed solely from internal dynamics. These findings suggest that reasoning in LLMs is fundamentally governed by geometric and informational constraints, offering a complementary framework to benchmark-centric assessment. --- ### Forecasting Source Stability in Scientific Experiments using Temporal Learning Models: A Case Study from Tritium Monitoring Authors: Nicholas Tan Jerome, Nadia Aouadi, Christoph Koehler, Suren Chilingaryan, Andreas Kopmann Published: 2026-05-02 Categories: physics.ins-det, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.08140v1 The Karlsruhe Tritium Neutrino Experiment (KATRIN) aims to measure the absolute neutrino mass with unprecedented sensitivity, requiring precise monitoring of the windowless gaseous tritium source, where tritium beta decay occurs. To track variations of the source activity, beta-induced X-ray spectroscopy provides real-time diagnostics. However, traditional drift detection methods struggle with the infrequent and transient nature of instability events in gaseous tritium. This study bridges the gap between state-of-the-art time-series forecasting models and real-world experimental applications by leveraging deep learning to predict the time to stability after instabilities. Unlike standard benchmarking approaches that emphasize algorithmic performance on fixed datasets, we apply forecasting models -- including LSTM, N-BEATS, TFT, NHITS, DLinear, NLinear, TSMixer, and Chronos-LLM -- to complex, large-scale experimental data. Our findings highlight two challenges: learning from sparse instability events and forecasting long time horizons (i.e., predicting hundreds of future points), both of which are ongoing challenges in time-series forecasting and remain active areas of research. This prediction task has direct experimental value by enabling better scheduling and maintenance planning. A reliable forecast of stability time allows for more efficient measurement and task management during stabilization periods. Through model selection, we identified N-BEATS as the top performer, excelling in accuracy and repeatability, demonstrating that deep learning can optimize large-scale physics experiments. --- ### Intelligent Autonomous Orchestration for Distributed Cloud Resources using Complex-Stability Analysis Authors: Gopal Krishna Shyam, Priyanka Bharti Published: 2026-05-02 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2605.08139v1 In modern distributed cloud environments, efficient resource allocation is required as traditional scaling mechanisms are often subject to cloud thrashing due to network-induced latencies. In this paper, we propose C-SAS (Complex-Stability Aware Scaling), an intelligent autonomous orchestration framework that leverages complex analytic methods to achieve system-wide equilibrium. In contrast to heuristic-based models, C-SAS acts as a stability-aware agent, converting telemetry noise into a deterministic "Safety Envelope" on the $s$-plane using the Argument Principle and Rouché's Theorem. The algorithm smartly suppresses oscillatory scaling operations that would otherwise degrade performance, by computing a real-time Analytic Stability Index (ASI). The experimental results show that C-SAS reduces VM flapping by 94\%, and achieves 96\% resource efficiency, significantly outperforming standard PID and ML-based autonomous agents. Our results suggest that future resilient autonomous cloud infrastructures will require AI-driven orchestrators with built-in formal stability constraints. --- ### DataArc-SynData-Toolkit: A Unified Closed-Loop Framework for Multi-Path, Multimodal, and Multilingual Data Synthesis Authors: Zhichao Shi, Cehao Yang, Hao Zhou, Xiaojun Wu, Huajie Li, Xuhui Jiang, Chengjin Xu, Yuanzhuo Wang, Jian Guo Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.08138v1 Synthetic data has emerged as a crucial solution to the data scarcity bottleneck in large language models (LLMs), particularly for specialized domains and low-resource languages. However, the broader adoption of existing synthetic data tools is severely hindered by convoluted workflows, fragmented data standards, and limited scalability across modalities. To address these limitations, we develop DataArc-SynData-Toolkit, an open-source framework featuring: (1) a configuration-driven, end-to-end pipeline equipped with an intuitive visual interface and simplified CLI for exceptional usability; (2) a unified, quality-controllable synthesis paradigm that standardizes multi-source data generation to ensure high reusability; and (3) a highly modular architecture designed for seamless multimodal, multilingual, and multi-task adaptation. We apply the toolkit in multiple application scenarios. Experimental results demonstrate that our toolkit achieves an optimal balance between generation efficiency and data quality. By offering an end-to-end and visually interactive pipeline, DataArc-SynData-Toolkit significantly lowers the technical barrier to synthetic data generation and subsequent model training, accelerating its practical deployment in real-world applications. --- ### Weight Pruning Amplifies Bias: A Multi-Method Study of Compressed LLMs for Edge AI Authors: Plawan Kumar Rath, Rahul Maliakkal Published: 2026-05-02 Categories: cs.LG, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2605.08137v1 Weight pruning is widely advocated for deploying Large Language Models on resource-constrained IoT and edge devices, yet its impact on model fairness remains poorly understood. We conduct a controlled empirical study of three instruction-tuned models (Gemma-2-9b-it, Mistral-7B-Instruct-v0.3, Phi-3.5-mini-instruct) across three pruning methods (Random, Magnitude, Wanda) at four sparsity levels (10-70%) on 12,148 BBQ bias benchmark items with 5 random seeds, totaling 2,368,860 inference records. Our results reveal a Smart Pruning Paradox: activation-aware pruning (Wanda) preserves perplexity nearly perfectly (just 3.5% increase at 50% sparsity for Mistral-7B), yet produces the highest bias amplification, with Stereotype Reliance Score increasing 83.7% and 47-59% of previously unbiased items developing new stereotypical behaviors at 70% sparsity. Random pruning destroys language capability entirely (perplexity exceeding $10^4$ and reaching $10^8$) but produces only random-chance bias. We further show that unstructured pruning provides zero storage savings and zero inference latency reduction on real edge hardware, undermining the primary motivation for its use in IoT deployment. Of 180 dense-vs-pruned comparisons, 141 (78.3%) are significant ($p < 0.05$) with mean $|h| = 0.305$. Published quantization studies report up to 21% of responses flipping between biased and unbiased states; our pruning results show transition rates nearly three times higher (47-59%), suggesting pruning poses a categorically greater risk to alignment than quantization. These findings demonstrate that perplexity-based evaluation provides false assurance of behavioral equivalence, and that IoT deployment pipelines require bias-aware validation before deploying pruned models at the edge. --- ### State Representation and Termination for Recursive Reasoning Systems Authors: Debashis Guha, Amritendu Mukherjee, Sanjay Kukreja, Tarun Kumar Published: 2026-05-02 Categories: cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2605.06690v1 Recursive reasoning systems alternate between acquiring new evidence and refining an accumulated understanding. Two design choices are typically left implicit: how to represent the evolving reasoning state, and when to stop iterating. This paper addresses both. We represent the reasoning state as an epistemic state graph encoding extracted claims, evidential relations, open questions, and confidence weights. We define the order-gap as the distance between the states reached by expand-then-consolidate versus consolidate-then-expand; a small order-gap suggests that the two orderings agree and further iteration is unlikely to help. Our main result gives a necessary and sufficient condition for the linearised order-gap to be non-degenerate near the fixed point, showing when the criterion is informative rather than algebraically vacuous. This is a local condition, not a global convergence guarantee. We apply the framework to recursive reasoning systems and sketch its application to agent loops, tree-of-thought reasoning, theorem proving, and continual learning. --- ### Dynamic Graph with Similarity-Aware Attention Graph Neural Network for Recommender Systems Authors: Aadarsh Senapati, Neha Kujur, Vivek Yelleti Published: 2026-05-02 Categories: cs.IR, cs.LG, cs.SI arXiv: https://arxiv.org/abs/2605.05238v1 Recommender systems are essential components of modern online platforms which presents personalized content in various domain. The traditional collaborative filtering methods depends on static user-item interaction graphs and a limited subset of similarity measures which fail to capture the changing nature of preferences of an individual. Recent graph neural network (GNN) based approaches focus on user-item bipartite graphs which do not use explicit user-user relational modelling and dynamic graph evolution during training. To address these limitations, this paper proposes a Dynamic Graph SimilarityAware Attention Graph Neural Network (DG-SA-GNN) framework that integrates dynamic user similarity graph construction with multi-similarity propagation and attention-based aggregation. The proposed architecture constructs four parallel user similarity graphs using Cosine, Jaccard, Discounted Pearson Correlation Coefficient (Discount PCC), and IPIJ similarity functions, each processed by a dedicated UserGNN module. A Graph Transformer fuses the four graph views, and a CrossAttention module refines user embeddings through interaction with item embeddings. Crucially, the graphs are reconstructed at scheduled epochs during training, enabling the model to adapt to the learned embedding space constituting the dynamic graph component. Mini-batch training with hard negative sampling improves scalability and convergence. Experiments on the MovieLens100K benchmark demonstrate that DG-SA-GNN achieves a Recall@20 of 0.162 and NDCG@20 of 0.065 which is better than the LightGCN baseline in recall. The results validate that dynamic multi-similarity graph construction coupled with attention-based fusion which produce recommendation performance --- ### Regularized Centered Emphatic Temporal Difference Learning Authors: Xingguo Chen, Chaohui Wu, Jinguo Ye, Chao Li, Shangdong Yang, Guang Yang, Tianyu Liang, Wenhao Wang Published: 2026-05-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.04100v1 Off-policy temporal-difference (TD) learning with function approximation faces a structural tradeoff among stability, projection geometry, and variance control. Emphatic TD (ETD) improves the off-policy projection geometry through follow-on emphasis, but the follow-on trace can have high variance. We revisit this tradeoff through Bellman-error centering. Although centering naturally removes a common drift term from TD errors, we show that a naive centered emphatic extension introduces an auxiliary coupling that can destroy the positive-definiteness of the ETD key matrix. We propose \emph{Regularized Emphatic Temporal-Difference Learning} (RETD), which preserves the follow-on trace and regularizes only the auxiliary centering recursion, corresponding to lifting the lower-right block of the coupled key matrix from \(1\) to \(1+c\). We derive the RETD core matrix, prove convergence under a conservative sufficient regularization condition, and evaluate the method on diagnostic linear off-policy prediction tasks. The experiments show that RETD avoids the instability of naive centered emphatic learning, preserves favorable emphatic geometry, and exhibits a robust intermediate regime for the regularization parameter \(c\) across the diagnostics. --- ### Tracing the Dynamics of Refusal: Exploiting Latent Refusal Trajectories for Robust Jailbreak Detection Authors: Xulin Hu, Che Wang, Wei Yang Bryan Lim, Jianbo Gao, Zhong Chen Published: 2026-05-02 Categories: cs.CR, cs.AI, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2605.02958v2 Representation Engineering analyses often characterize refusal using static directions extracted from terminal or pooled representations. We ask whether this view misses how refusal is constructed across layer-token positions. Using causal tracing, we identify a \textit{Refusal Trajectory}: a sparse upstream activation pattern that often persists even when attacks such as GCG suppress terminal refusal signals. Based on this observation, we propose SALO (Sparse Activation Localization Operator), a lightweight white-box detector that operates on raw hidden-state volumes from a selected layer window. Across Qwen, Llama, and Mistral models, SALO improves jailbreak detection on several attack families under a fixed XSTest-calibrated operating point. We further analyze static RepE-style baselines, ROI sensitivity, adaptive GCG attacks, and encoded-input boundary cases, clarifying both the promise and limitations of refusal-trajectory monitoring. --- ### EFGPP: Exploratory framework for genotype-phenotype prediction Authors: Muhammad Muneeb, David B. Ascher Published: 2026-05-02 Categories: q-bio.GN, cs.LG arXiv: https://arxiv.org/abs/2605.02954v1 Predicting complex human traits from genetic data is challenging because different genetic, clinical, and molecular data sources often contain different parts of the signal. Here, we present EFGPP, a reproducible framework for generating, ranking, and combining multiple types of data for genotype-to-phenotype prediction. We applied EFGPP to migraine prediction using UK Biobank data from 733 individuals. The framework combined genotype-derived features, principal components, clinical and metabolomic covariates, and polygenic risk scores generated from migraine and depression GWAS using PLINK, PRSice-2, AnnoPred, and LDAK-GWAS. The best single data type achieved a test AUC of 0.644, while combining multiple data types improved performance to 0.688 using migraine-focused inputs and 0.663 using cross-trait depression-derived inputs. Genetic features alone did not outperform the covariates-only baseline, but genotype-derived features performed better than PRS alone, and depression-derived PRS showed useful predictive signal. Overall, EFGPP provides a practical proof-of-concept framework for prioritising and integrating heterogeneous genetic data sources for complex phenotype prediction. --- ### AI Alignment via Incentives and Correction Authors: Rohit Agarwal, Joshua Lin, Mark Braverman, Elad Hazan Published: 2026-05-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.01643v3 We study AI alignment through the lens of law-and-economics models of deterrence and enforcement. In these models, misconduct is not treated as an external failure, but as a strategic response to incentives: an actor weighs the gain from violation against the probability of detection and the severity of punishment. We argue that the same logic arises naturally in agentic AI pipelines. A solver may benefit from producing a persuasive but incorrect answer, hiding uncertainty, or exploiting spurious shortcuts, while an auditor or verifier must decide whether costly monitoring is worthwhile. Alignment is therefore a fixed-point problem: stronger penalties may deter solver misbehavior, but they can also reduce the auditor's incentive to inspect, since auditing then mainly incurs cost on a population that appears increasingly aligned. This perspective also changes what should count as a post-training signal. Standard feedback often attaches reward to the final answer alone, but a solver-auditor pipeline exposes the full correction event: whether the solver erred, whether the auditor inspected, whether the error was caught, and whether oversight incentives remained active. We formalize this interaction in a two-agent model in which a principal chooses rewards over joint correction outcomes, inducing both solver behavior and auditor monitoring. Reward design is therefore a bilevel optimization problem: rewards are judged not by their immediate semantic meaning, but by the behavioral equilibrium they induce. We propose a bandit-based outer-loop procedure for searching over reward profiles using noisy interaction feedback. Experiments on an LLM coding pipeline show that adaptive reward profiles can maintain useful oversight pressure and improve principal-aligned outcomes relative to static hand-designed rewards, including a substantial reduction in hallucinated incorrect attempts. --- ### Adaptive Pluralistic Alignment: A pipeline for dynamic artificial democracy Authors: Rachel Freedman Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.01642v2 Prevailing alignment methods target a fixed set of preferences and therefore risk forcing value lock-in as societal norms evolve over time. We introduce Adaptive Pluralistic Alignment (APA), a modular pipeline for updating pluralistically aligned AI systems to track evolving values and avoid value lock-in without repeating costly pretraining or large-scale data collection. APA has three stages: (1) learning compact personalized reward models via low-rank reward basis decomposition, (2) using these models as a jury that collectively selects among candidate outputs through social-choice-theoretic voting, and (3) efficiently adapting the jury over time by fitting new annotator weights over the fixed reward bases as values shift. The resulting system is efficient, explainable, steerable, and modular. We implement a proof-of-concept instantiation using the PRISM multi-user alignment dataset and simulated historical annotators, and provide preliminary analysis showing that jury composition and the choice of voting rule can substantially affect outcomes, particularly when jury preferences are heterogeneous. We provide full code and resulting preference datasets at https://github.com/RachelFreedman/apa. --- ### Prescriptive Scaling Laws for Data Constrained Training Authors: Justin Lovelace, Christian Belardi, Srivatsa Kundurthy, Shriya Sudhakar, Kilian Q. Weinberger Published: 2026-05-02 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2605.01640v1 Training compute is increasingly outpacing the availability of high-quality data. This shifts the central challenge from optimal compute allocation to extracting maximum value from limited data. The widely adopted Chinchilla scaling law assumes every training token is unique. This limits its ability to guide pretraining decisions in data-constrained regimes. We model the excess loss under repetition with a simple additive overfitting penalty and find that it accurately describes model behavior. Our scaling law yields qualitatively new compute-optimal allocation advice. Beyond a point, further repetition is counterproductive and compute is better spent on model capacity. We show that following our law's recommended configuration improves performance in data-constrained regimes. Finally, because our one-parameter form isolates overfitting in a single coefficient, it enables direct comparison across training configurations. As a case study, we show that strong weight decay ($λ=1.0$) reduces this coefficient by approximately 70%, providing a scaling-law explanation for recent findings that optimal weight decay in data-constrained regimes is an order of magnitude larger than standard practice. --- ### Omni-Fake: Benchmarking Unified Multimodal Social Media Deepfake Detection Authors: Tianxiao Li, Zhenglin Huang, Haiquan Wen, Yiwei He, Xinze Li, Bingyu Zhu, Wuhui Duan, Congang Chen, Zeyu Fu, Yi Dong, Baoyuan Wu, Jason Li et al. Published: 2026-05-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.01638v1 Multimodal deepfakes are proliferating on social media and threaten authenticity, information integrity, and digital forensics. Existing benchmarks are constrained by their single-modality scope, simplified manipulations, or unrealistic distributions, which limit their ability to assess real-world robustness. To address these limitations, we present Omni-Fake, a unified omni-dataset for comprehensive multimodal deepfake detection in social-media settings. It comprises Omni-Fake-Set, a large-scale, high-quality dataset with 1M+ samples, and Omni-Fake-OOD, an out-of-distribution benchmark with 200k+ samples intentionally excluded from training to evaluate generalization. Omni-Fake spans four modalities (image, audio, video, and audio-video talking head) and supports a joint detection-localization-explanation protocol. On top of Omni-Fake, we further propose Omni-Fake-R1, a reinforcement-learning-driven multimodal detector that adaptively integrates visual and auditory cues and outputs structured decisions, localization, and natural-language explanations. Extensive experiments show significant gains in detection accuracy, cross-modal generalization, and explainability over state-of-the-art baselines. Project page: https://tianxiao1201.github.io/omni-fake-project-page/ --- ### The Banach-Butterfly Invariant: Influence-Adaptive Walsh Geometry for Ternary Polynomial Threshold Functions Authors: Gorgi Pavlov Published: 2026-05-02 Categories: cs.LG, cs.CC, cs.DM, math.CO arXiv: https://arxiv.org/abs/2605.01637v1 We introduce the Banach-Butterfly Invariant (BBT), an influence-adaptive Banach geometry on the Walsh-Hadamard butterfly factorization. For a Boolean function $f:\{-1,+1\}^n\to\{-1,+1\}$ with coordinate influences $\mathrm{Inf}_\ell(f)$, BBT assigns exponent $p_\ell = 1+\mathrm{Inf}_\ell(f)$ to butterfly layer $\ell$, yielding the contraction invariant $μ(f)=\prod_\ell 2^{-\mathrm{Inf}_\ell/(1+\mathrm{Inf}_\ell)}$. We prove a Jensen lower bound $\log_2μ(f) \ge -I(f)/(1+I(f)/n)$ and that $μ$ is strictly Schur-convex in the influence vector (modulo permutation), giving scaling classes $μ\sim 2^{-n/2}$ (parity), $2^{-Θ(\sqrt{n})}$ (majority), $2^{-1/2}$ (dictators). $\log_2μ$ is rational but not polynomial in the Fourier coefficients while $μ$ is algebraic, and $μ$ separates functions with identical total influence (122 pairs at $n=3$). Using the certified $n \le 4$ ternary Walsh-threshold universe from a companion synthesis manuscript as a finite testbed, we compute exact MILP minimum-support certificates for all 65,536 Boolean functions at $n=4$ (mean 6.42, max 9, all-odd by a parity argument) and on 10,000 of the 616,126 NPN-canonical representatives we enumerate at $n=5$ (matching OEIS A000370). Conditional Spearman $ρ(μ,|\mathrm{supp}|)$ at fixed total influence is $+0.571$ in the largest stratum at $n=4$ but reverses to $-0.38$ at $n=5$ under both function-uniform and NPN-canonical sampling: $μ$ is a valid Schur-convex concentration invariant, not a universal monotone predictor of minimum support across $n$. A companion application paper validates a real-valued WHT activation-energy proxy inspired by this theory on five pretrained LLMs at W2A16, cutting wikitext-2 perplexity by 15-58% versus vanilla auto-round; the transfer from Boolean theory to the real-valued proxy is qualitative, not formal. --- ### Chebyshev-Augmented One-Shot Transfer Learning for PINNs on Nonlinear Differential Equations Authors: Yiqi Rao, Pavlos Protopapas Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.01634v1 Physics-Informed Neural Networks (PINNs) offer a flexible paradigm for solving differential equations by embedding governing laws into the training objective. A persistent limitation is instance specificity: standard PINNs typically require retraining for each new forcing term, boundary/initial condition, or parameter setting. One-shot transfer learning (OTL) addresses this bottleneck for linear operators by freezing a pretrained latent representation and computing optimal output weights in closed form, but for nonlinear problems closed-form adaptation is generally unavailable because the loss is nonconvex in the output layer. In this paper we substantially broaden the class of nonlinearities amenable to one-shot PINN transfer by combining OTL with Chebyshev polynomial surrogates. We approximate general smooth weakly nonlinear terms by truncated Chebyshev expansions over a prescribed solution range, yielding a polynomial nonlinearity that can be handled by a perturbative decomposition into linear subproblems. A multi-head PINN learns a reusable latent space associated with the dominant linear operator; at test time, solutions to new instances are obtained via a sequence of closed-form linear solves in the output layer, without retraining the network body. We provide a unified derivation of the framework for ODEs and PDEs and demonstrate accuracy and fast online adaptation on nonlinear benchmarks, including non-polynomial and singular ODE nonlinearities as well as a reaction-diffusion PDE with saturating kinetics, demonstrating the method's utility in many-query regimes. --- ### Perturb and Correct: Post-Hoc Ensembles using Affine Redundancy Authors: Eleanor Quint Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.01632v1 Models that are indistinguishable on in-distribution data can behave very differently under distribution shift. We introduce Perturb-and-Correct (P&C), a post-hoc method for constructing epistemically diverse predictors from a single pretrained network. P&C applies random hidden layer perturbations with a least-squares correction in the subsequent affine layer, producing predictors that agree on calibration data while remaining free to disagree away from it. We analyze this mechanism through the post-correction residual and its first-order sensitivity: the residual is controlled near the calibration distribution by a leverage term, while corrected sensitivity grows as inputs deviate from the calibration geometry. Empirically, P&C achieves a strong ID/OOD tradeoff across MuJoCo dynamics prediction and CIFAR-10 OOD detection, matching or outperforming standard post-hoc baselines while requiring only a single pretrained model. Our findings highlight the potential in further exploiting overparameterization as a strength of deep learning models. --- ### Prosa: Rubric-Based Evaluation of LLMs on Real User Chats in Brazilian Portuguese Authors: Roseval Malaquias Junior, Giovana Kerche Bonás, Thales Sales Almeida, Hugo Abonizio, Thiago Laitz, Ramon Pires, Marcos Piau, Celio Larcher, Rodrigo Nogueira Published: 2026-05-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.01630v1 Rankings produced by holistic LLM-as-a-judge scoring are sensitive to the bias of the chosen judge model. We show that switching to binary rubric scoring with multi-judge filtering removes this sensitivity: decomposing the judgement matters more than the judge model itself. To support this claim, we introduce Prosa, the first real user multi-turn Brazilian Portuguese chat benchmark: 1,000 WildChat conversations scored by three judges from three model families on 16 models. Under filtered rubric scoring the three judges agree on every one of the 16 ranks, whereas under holistic scoring they agree on only 7 of 16. Additionally, the rubric filtering pipeline increases the average score gap between neighbouring models by 47%, thereby improving Prosa's discriminative power. Evaluating a new model on Prosa costs approximately $2.1 when using Gemini 3 Flash as the judge. We release the benchmark and the filtering code to ensure that future models can be assessed under identical conditions. These artifacts also make our rubric-based scoring method reusable beyond Prosa, supporting other open-ended evaluation settings. --- ### Self-Normalized Martingales and Uniform Regret Bounds for Linear Regression Authors: Fan Chen, Jian Qian, Alexander Rakhlin, Nikita Zhivotovskiy Published: 2026-05-02 Categories: stat.ML, cs.LG, math.ST arXiv: https://arxiv.org/abs/2605.01628v1 Self-normalized martingale inequalities lie at the heart of confidence ellipsoids for online least squares and, more broadly, many bandit and reinforcement-learning results. Yet existing vector and scalar results typically rely on bounded covariates and an explicit regularization matrix, producing bounds that are \emph{not scale-invariant}: although the self-normalized quantity is scale-invariant by definition, its standard upper bounds are not. We characterize when scale-invariant upper bounds on self-normalized martingales are possible. Without further assumptions, we prove that nontrivial scale-invariant bounds exist only in dimension $d=1$; moreover, in $d=1$ we obtain $O(\log T)$ scale-invariant self-normalized bounds without any assumptions on the covariates. In contrast, for $d>1$ we show that no nontrivial scale-invariant bound can hold in full generality. We then connect this dichotomy to \emph{doubly-uniform} regret in online linear regression (i.e., regret bounds that are simultaneously independent of the covariate scale and the comparator norm) and use it to resolve the open question of Gaillard, Gerchinovitz, Huard, and Stoltz, \emph{``Uniform regret bounds over $\mathbb{R}^d$ for the sequential linear regression problem with the square loss''} (ALT 2019): in $d=1$ we give an explicit algorithm with $O(\log T)$ doubly-uniform regret, whereas for $d>1$ sublinear doubly-uniform regret is impossible. Finally, under a natural \emph{smoothness} condition (bounded Radon--Nikodym derivatives of the conditional covariate laws with respect to a fixed base measure), we recover sublinear regret for $d>1$ without bounded covariates and derive a self-normalized concentration inequality free of the usual regularization penalties, yielding arguably a first natural scale-invariant bound for adaptive, non-i.i.d. vector martingales. --- ### Importance-Guided Basis Selection for Low-Rank Decomposition of Large Language Models Authors: Daniel Agyei Asante, Ernie Chang, Yang Li Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.01627v2 Low-rank decomposition is a compelling approach for compressing large language models, but its effectiveness hinges on selecting which singular-vector bases to retain for a target task. Existing methods such as Basel adapt singular-value coefficients on downstream data and prune bases with small re-learned magnitudes, a heuristic that can be misaligned with task performance because it ignores the local geometry of the loss landscape. We present Basis Selection with Importance (BSI), a principled low-rank compression framework that ranks and prunes bases by directly estimating the expected loss increase incurred when each basis is removed. BSI derives a derivative-based importance score from a second-order Taylor expansion of the task loss with respect to singular values, combining first-order sensitivity and second-order curvature to quantify pruning impact. To make this criterion practical for LLMs, we develop an efficient Hessian-diagonal estimator by adapting the Hutchinson randomized-probing method to loss curvature with symmetric parameter perturbations. We provide a comprehensive theoretical analysis, including loss-increase bounds under basis pruning, explicit propagation of Hessian-diagonal estimation error into these bounds, variance characterization tied to the Hessian spectrum, high-probability sample-complexity guarantees for achieving a target estimation accuracy, and guidance on perturbation intensity. Extensive experiments on mathematical reasoning benchmarks demonstrate that BSI consistently outperforms state-of-the-art low-rank decomposition baselines, with especially strong improvements under deep compression. --- ### PRIME: Protein Representation via Physics-Informed Multiscale Equivariant Hierarchies Authors: Viet Thanh Duy Nguyen, John K. Johnstone, Truong-Son Hy Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.01625v3 Proteins are inherently multiscale physical systems whose functional properties emerge from coordinated structural organization across multiple spatial resolutions, ranging from atomic interactions to global fold topology. However, existing protein representation learning methods typically operate at a single structural level or treat different sources of structural information as parallel modalities, without explicitly modeling their hierarchical relationships. We introduce PRIME (Protein Representation via Physics-Informed Multiscale Equivariant Hierarchies), a unified framework that models proteins as a nested family of five physically grounded structural graphs spanning surface, atomic, residue, secondary-structure, and protein levels. Adjacent levels are connected through deterministic, physics-informed assignment operators, enabling bidirectional information exchange via bottom-up aggregation and top-down contextual refinement. Experiments on standard protein representation learning benchmarks demonstrate strong and competitive performance across diverse tasks, with particularly notable gains on the Fold Classification benchmark, where PRIME outperforms the strongest geometric GNN baseline by margins of 13.80 and 18.30 points on the harder Superfamily and Fold splits, and achieves a state-of-the-art accuracy of 84.10\% on Reaction Class prediction, surpassing all baseline methods, including ESM. Ablation studies confirm that each structural level contributes complementary and non-redundant information, and adaptive cross-attention analysis reveals that PRIME autonomously identifies the most task-relevant structural resolutions at prediction time. Our source code is publicly available at https://github.com/HySonLab/PRIME --- ### Persistent Homology of Time Series through Complex Networks Authors: İsmail Güzel Published: 2026-05-02 Categories: math.AT, cs.LG, stat.AP, stat.ML arXiv: https://arxiv.org/abs/2605.01624v1 We present a unified pipeline for univariate time series classification via complex networks and persistent homology. A time series is mapped to a graph through one of five constructions across three families (visibility (natural and horizontal visibility graphs), transition, and proximity) and the graph is converted to a dissimilarity matrix from which a Vietoris-Rips filtration yields persistence diagrams. These diagrams are vectorized into fixed-length features through persistence landscapes and topological summary statistics. By standardizing the downstream processing, differences in classification performance are attributable to the network construction and distance metric alone. Experiments on twelve UCR benchmarks show that (i) no single construction dominates: the optimal graph type depends on the signal's discriminative structure; (ii) the graph distance metric is a first-order design choice, with diffusion distance uniformly outperforming shortest-path alternatives; and (iii) persistence-based features degrade gracefully under noise, consistent with the classical stability theorem of persistent homology. --- ### Learning Behavioral Signals from Encrypted Smartphone Network Traffic Authors: Rameen Mahmood, Omar El Shahawy, Souptik Barua, Zachary Beattie, Jeffrey Kaye, Xuhai "Orson'' Xu, Chao-Yi Wu, Danny Yuxing Huang Published: 2026-05-02 Categories: cs.LG, cs.AI, cs.CY, cs.NI arXiv: https://arxiv.org/abs/2605.01616v2 Human behavior is challenging to measure continuously at scale, yet traces of daily routines and well-being may be reflected in interactions with personal devices. We investigate whether encrypted smartphone network traffic can serve as a passive sensing signal for behavioral states related to sleep disturbance, stress, and loneliness. To capture both population-level patterns and individual-specific behavior, we employ a transformer-based model with user-specific adapters that learns representations of network activity while accounting for personal baselines and deviations from them. To improve interpretability, we further analyze these representations using sparse representation learning to identify latent behavioral features associated with distinct activity patterns. We relate the resulting features to sleep disturbance, stress, and loneliness using generalized estimating equations with Mundlak decomposition, enabling separation of stable between-person differences from within-person changes over time. Our analysis reveals that the three outcomes are characterized by different temporal dynamics: stress is predominantly associated with persistent between-person variation, loneliness is more strongly linked to within-person fluctuations, and sleep disturbance reflects a combination of both. Importantly, these within-person behavioral signals are not recovered by conventional handcrafted network-traffic features, highlighting the advantages of learned representations for longitudinal behavioral modeling. Overall, our findings demonstrate that encrypted network traffic contains interpretable behavioral information and can support passive, scalable monitoring of behavioral dynamics, particularly changes relative to an individual's typical pattern of activity. --- ### The Case for ESM3 as a General-Purpose AI Model with Systemic Risk Under the EU AI Act Authors: Taro Qureshi, Jacob Griffith, Koen Holtman, Marcel Mir Teijeiro, Ze Shen Chin, Rokas Gipiškis Published: 2026-05-02 Categories: cs.CY, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2605.01611v1 Due to ambiguity in the wording of the EU AI Act, we examine the question of to what extent frontier biological foundation models such as ESM3 are subject to obligations for general-purpose AI models with systemic risk under the EU AI Act. In this paper, we map ESM3 to the biorisk chain, and conclude that it would be desirable if the providers of ESM3 and similar biological models were subject to these obligations, which would require them to assess and mitigate dual-use risks from their models. We then perform an analysis, comparing the attributes of ESM3 to the classification criteria in the AI Act and the supporting material. We conclude that at this time, ESM3 does not appear to be meaningfully regulated by the Act. We then propose remedies to correct the situation. --- ### Less Interaction But More Explanation: A Communication Perspective on Agentic AI Interfaces Authors: Eunchae Jang, S. Shyam Sundar Published: 2026-05-02 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2605.01610v1 AI systems have long been expected to interact with users, answering questions, generating content, and continuing (social) conversations. Agentic AI, however, breaks from this expectation, as its primary objective is workflow execution on behalf of the users. If a system becomes more agentic, do users need less interaction with the system? Our answer is: less routine back-and-forth, but more communication for oversight and explanation, as agentic AI proactively acts, not just responds. Grounded in a communication perspective, we discuss how users perceive the communicative roles of AI systems (whether as the source of actions or merely a channel), and how this can shape trust. Because agentic AI can play multiple communicative roles, it can complicate this source perception and introduce potential risks. To address this, we propose three types of explanations that agentic AI needs to incorporate (action-process, uncertainty, and coordination), and suggest that customization affordances that allow users to decide when and which explanations they see may be key to preserving human agency as AI autonomy increases. --- ### Concepts Whisper: Spectral Anti-Concentration and the Dual Geometry of Transformer Representations Authors: Pratyush Acharya, Nuraj Rimal, Habish Dhakal Published: 2026-05-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.01609v2 We find that transformer concept representations systematically anti-concentrate in the spectral tail of the unembedding covariance, encoding word-level concepts in low-variance directions across a 17-model core suite and an expanded set of 22 semantic concept categories, with convergent replications from three independent extraction methods. Residual-stream difference-of-means vectors anti-concentrate in all 17 models (model-level one-sample t-test, p = 3.8e-9), and remain more tail-aligned than norm-matched random directions in 13 of 17; convergent support comes from sparse autoencoder (SAE) features (p = 4.5e-19 across concepts within a model) and linear probes on Llama and Qwen. We identify a dual geometry: activation-space concept directions anti-concentrate while static unembedding-row contrasts concentrate in high-variance directions (p < 10^-4). This investigation arose from testing whether the causal inner product of Park et al. (2024) aids cross-lingual concept transport; a matched-spectrum randomization across 17 models and four language pairs finds no evidence that Whitened Causal Alignment improves over spectral regularization alone (p = 0.95). Split-injection interventions, restricted to steering strengths at which both arms remain interpretable, show the predicted interference asymmetry in four of five models (paired Cohen's d_z up to 1.19) with no significant reversal inside that regime, and POS-tag probing across eight models shows syntax preferentially encoded in the high-variance subspace in six of eight architectures, with a significant reversal in the Qwen 2.5 family. These results suggest transformers rotate semantic content into spectrally quiet regions during contextualized processing, where, in some architectures, interventions may reduce grammatical disruption relative to high-variance steering. --- ### Concepts Whisper While Syntax Shouts: Spectral Anti-Concentration and the Dual Geometry of Transformer Representations Authors: Pratyush Acharya, Nuraj Rimal, Habish Dhakal Published: 2026-05-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.01609v1 We test whether the causal inner product of \citet{park2024linear} -- defined by the unembedding covariance $Σ$ -- enables cross-lingual concept transport. Across 17 models and 4 language pairs, a matched-spectrum randomization test finds that Whitened Causal Alignment is indistinguishable from spectral regularization alone ($p = 0.95$). However, this failure reveals a broader phenomenon: anti-concentration is observed in residual-stream difference-of-means vectors across five architecture families ($p < 10^{-33}$) and supported by SAE features (e.g., $p = 4.5 \times 10^{-19}$) and linear probes on Gemma and Llama. We discover a \emph{dual geometry}: activation-space concept directions anti-concentrate in the spectral tail, while static unembedding-row contrasts \emph{concentrate} in high-variance directions ($p < 10^{-4}$). Split-injection causal interventions support the functional basis on Gemma and Llama (Cohen's $d$ up to $1.80$), and POS-tag probing across 8 models shows syntax preferentially encodes in the high-variance subspace in 6 of 8 architectures ($p < 0.013$), with the Qwen~2.5 family showing a significant reversal consistent with architecture-specific spectral structure. These results suggest transformers may rotate semantic content into spectrally quiet regions during contextualized processing, encoding concepts where they can be manipulated with reduced grammatical disruption. --- ### Why Model Selection Fails in Time Series Forecasting: An Empirical Study of Instability Across Data Regimes Authors: Tahir Cetin Akinci, Alfredo A. Martinez-Morales Published: 2026-05-02 Categories: eess.SP, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2605.01608v1 Time series forecasting models often exhibit inconsistent performance across datasets with varying statistical and structural properties. Despite the wide range of available forecasting techniques, it remains unclear whether model selection can be reliably guided by simple data characteristics. This paper investigates why rule-based model selection fails in time series forecasting by analyzing the relationship between data-regime descriptors and model performance. A descriptor-based framework is introduced to characterize time series using measurable properties, including trend strength, seasonality, noise level, and temporal dependence. Based on these descriptors, a rule-based selection mechanism is formulated to map data regimes to candidate forecasting models. The approach is evaluated on multiple real-world datasets across different domains and forecasting horizons. The results show that rule-based model selection achieves low accuracy, with correct model identification occurring in only a small fraction of cases. Significant discrepancies are observed between recommended and empirically optimal models, particularly in noisy and mixed regimes. Further analysis reveals that model performance is highly sensitive to both dataset characteristics and forecasting horizon, resulting in substantial ranking instability across scenarios. These findings explain why simple heuristic rules fail to generalize and demonstrate that forecasting performance cannot be reliably predicted using static, descriptor-based approaches. This study provides empirical evidence that model selection in time series forecasting is inherently context-dependent and highlights the need for more adaptive, data-driven strategies. --- ### Where Do Prompt Perturbations Break Generation? A Segment-Level View of Robustness in LoRA-Tuned Language Models Authors: Zhuoyun Li, Boxuan Wang, Jinwei Hu, Zhenglin Huang, Qisong He, Xinmiao Huang, Guangliang Cheng, Xiaowei Huang, Yi Dong Published: 2026-05-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2605.01605v1 Large language models are sensitive to minor prompt perturbations, yet existing robustness methods usually enforce consistency at the whole-sequence level. This holistic view can hide an important failure mode: a perturbed response may remain globally similar to the clean one while drifting on a critical entity, relation, or conclusion. We introduce S$^2$R$^2$, a segment-level framework for robust LoRA fine-tuning. S$^2$R$^2$ decomposes clean and perturbed generations into semantic segments, aligns them with an optimal-transport objective, and penalises the segments with the largest meaning drift. To connect this output-side objective with model adaptation, we add an adapter-stability regulariser motivated by segment-level attention reallocation, using LoRA norm control as a tractable proxy for limiting perturbation-amplified evidence shifts. A PAC-Bayesian complexity view further explains why controlling adapter growth may support transfer beyond observed perturbations. Experiments on summarisation benchmarks show that S$^2$R$^2$ improves robustness under typographical noise, deletion, synonym replacement, and paraphrasing, while maintaining competitive clean performance and stronger cross-dataset transfer than consistency-based baselines. --- ### Evaluating Agentic AI in the Wild: Failure Modes, Drift Patterns, and a Production Evaluation Framework Authors: Mukund Pandey Published: 2026-05-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.01604v1 Existing evaluation frameworks for large language models -- including HELM, MT-Bench, AgentBench, and BIG-bench -- are designed for controlled, single-session, lab-scale settings. They do not address the evaluation challenges that emerge when agentic AI systems operate continuously in production: compounding decision errors, tool failure cascades, non-deterministic output drift, and the absence of ground truth for long-horizon tasks. This paper makes three contributions. First, we present a taxonomy of seven failure modes unique to production agentic systems, each grounded in observations from systems operating at billion-event scale. Second, we demonstrate empirically where standard metrics -- ROUGE, BERTScore, accuracy/AUC, and the agentic benchmarks above -- fail to detect each failure mode. Third, we propose PAEF (Production Agentic Evaluation Framework), a five-dimension evaluation framework with an open-source reference implementation, designed for continuous evaluation on production traffic rather than episodic benchmark runs. Our analysis shows that standard metrics fail to detect four of the seven failure modes entirely and detect three others only after a lag of multiple evaluation cycles. --- ### Fine-Tuning Pre-Trained Code Models for AI-Generated Code Detection Authors: Jany-Gabriel Ispas, Sergiu Nisioi Published: 2026-05-02 Categories: cs.CL arXiv: https://arxiv.org/abs/2605.01596v1 This paper describes the system submitted by team \textbf{Archaeology} to SemEval-2026 Task~13 on AI-generated code detection. The shared task consists of three subtasks; we participate in Subtask-A (binary classification: human-written vs.\ AI-generated code) and Subtask-B (11-class attribution of the generating model). Starting from a TF-IDF and Logistic Regression baseline, we fine-tune four pre-trained code models (CodeBERT, GraphCodeBERT, UniXcoder, and CodeT5+) with separate strategies for each subtask. For Subtask-A, we use leave-one-language-out cross-validation, code augmentation, chunked inference with trimmed-mean aggregation, and threshold calibration on a difficult dataset. For Subtask-B, we use sandwich token packing, class-balanced loss, and multi-seed ensembling with test-time augmentation. Our best submissions obtain macro-F1 scores of 0.737 on Subtask-A (6th/81 teams) and 0.422 on Subtask-B (7th/34 teams). --- ### Led to Mislead: Adversarial Content Injection for Attacks on Neural Ranking Models Authors: Amin Bigdeli, Amir Khosrojerdi, Radin Hamidi Rad, Morteza Zihayat, Charles L. A. Clarke, Ebrahim Bagheri Published: 2026-05-02 Categories: cs.IR, cs.CL arXiv: https://arxiv.org/abs/2605.01591v1 Neural Ranking Models (NRMs) are central to modern information retrieval but remain highly vulnerable to adversarial manipulation. Existing attacks often rely on heuristics or surrogate models, limiting effectiveness and transferability. We propose CRAFT, a supervised framework for black-box adversarial rank attacks powered by large language models (LLMs). CRAFT operates in three stages: adversarial dataset generation via retrieval-augmented generation and self-refinement, supervised fine-tuning on curated adversarial examples, and preference-guided optimization to align generations with rank-promotion objectives. Extensive experiments on the MS MARCO passage dataset, TREC Deep Learning 2019, and TREC Deep Learning 2020 benchmarks show that CRAFT significantly outperforms state-of-the-art baselines, achieving higher promotion rates and rank boosts while preserving fluency and semantic fidelity. Moreover, CRAFT transfers effectively across diverse ranking architectures, including cross-encoder, embedding-based, and LLM-based rankers, underscoring vulnerabilities in real-world retrieval systems. This work provides a principled framework for studying adversarial threats in NRMs, underscores the risks of generative AI in rank manipulation, and provides a foundation for developing more robust retrieval systems. To support reproducibility, we publicly release our source code, trained models, and prompt templates. --- ### KG-First, LLM-Fallback: A Hybrid Microservice for Grounded Skill Search and Explanation Authors: Ngoc Luyen Le, Marie-Hélène Abel, Bertrand Laforge Published: 2026-05-02 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2605.01582v1 Authoritative competency frameworks such as ESCO, ROME, and O*NET are essential for aligning education with labor market needs, yet their technical complexity and structural heterogeneity hinder practical adoption by educators. This paper introduces SkillGraph-Service, an interoperable microservice designed to bridge this gap by unifying these resources into a provenance-preserving Knowledge Graph (KG). Adopting a KG-first, LLM-fallback architecture, the system combines symbolic rigor with sub-symbolic flexibility. It implements a lightweight hybrid retrieval engine (fusing SQLite FTS5 and HNSW vector search) to handle the vocabulary mismatch in educator queries, and utilizes Large Language Models (LLMs) strictly for constrained ranking and audience-aware explanation. Empirical evaluation on a multilingual dataset reveals that the proposed hybrid strategy achieves superior retrieval effectiveness (nDCG@5>0.94) with sub-200 ms latency, rendering computationally expensive cross-encoder re-ranking may be unnecessary for this domain. Furthermore, an analysis of generated explanations highlights a trade-off between fluency and faithfulness: while JSON-constrained LLMs ensure high citation precision, deterministic templates remain the most reliable method for maximizing evidence coverage. The resulting architecture offers a practical, scalable, and auditable solution for integrating complex skill data into digital learning ecosystems. --- ### Model Merging: Foundations and Algorithms Authors: Donato Crisostomi Published: 2026-05-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2605.01580v1 Modern deep learning usually treats models as separate artifacts: trained independently, specialized for particular purposes, and replaced when improved versions appear. This thesis studies model merging as an alternative paradigm: combining independently trained neural networks directly in weight space, with little or no optimization and without requiring access to the original training data. The thesis considers two main regimes. In the single-task setting, where models share an objective but differ in initialization, we introduce C$^2$M$^3$, a cycle-consistent merging algorithm based on Frank-Wolfe optimization. C$^2$M$^3$ aligns multiple networks into a shared, reference-free parameter space, making weight averaging meaningful without privileging any individual model. In the multi-task setting, where models are fine-tuned for different downstream tasks from a common pretrained initialization, we first develop a theoretical account of task vectors as approximate gradients. This explains both the effectiveness and the limitations of task arithmetic. Building on this view, we show that task vectors inherit the low-rank structure of gradients and introduce Task Singular Vectors (TSV), a decomposition that enables compression and interference reduction through TSV-Merge. We then present MASS, an input-adaptive routing method that uses TSV geometry to select task-relevant subspaces at inference time. Finally, we introduce MERGE$^3$, an evolutionary merging framework that uses Item Response Theory to reduce evaluation costs by up to 50$\times$ while preserving solution quality. Together, these contributions provide theoretical and algorithmic foundations for model merging, supporting a paradigm in which learned capabilities can be composed, reused, and extended across models. --- ### Minimum Specification Perturbation: Robustness as Distance-to-Falsification in Causal Inference Authors: Hoang Dang, Luan Pham, Minh Nguyen Published: 2026-05-02 Categories: stat.ME, cs.LG arXiv: https://arxiv.org/abs/2605.01579v1 Empirical causal claims depend on many analyst decisions, from selecting covariates to choosing estimators. Existing robustness tools summarize how results vary across these choices, but, to the best of our knowledge, do not answer: \textbf{How many analyst decisions must change to reach a specification, which is a set of choices, whose confidence interval (CI) contains zero?} We introduce \emph{Minimum Specification Perturbation (MSP)}, the smallest number of changes. MSP is small under the null, grows with effect strength and captures distance-to-falsification information that dispersion-based summaries cannot report; when making decisions under weak effects, an MSP-based rule yields lower false-positive rates than dispersion-based rules. We show that Fragility Index and MSP measure orthogonal vulnerabilities: fragility to influential observations need not imply fragility to specification choices. On the LaLonde benchmark, MSP = 1 implies that one decision change makes the CI contain zero. We further provide exact permutation calibration under randomization and characterize computation, showing tractable cases under additive structure and NP-hardness in general. --- ### Hybrid Quantum Reinforcement Learning with QAOA for Improved Vehicle Routing Optimization Authors: T. Satyanarayana Murthy, B. Swathi Sowmya, Santhosh Voruganti, Sai Varshini Giridi, Chaitanyya Pratap Agarwal, Vanteddu Akshitha Published: 2026-05-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2605.01574v1 Vehicle Routing Problem (VRP) is one of the most complex NP-hard combinatorial optimization problem in transportation and logistics that requires a dynamic solution approach. In this paper we present a new hybrid approach that combines the Quantum Approximate Optimization Algorithm (QAOA) into the QRL policy network, instead of the usual variational layers, QAOA mixing and cost Hamiltonian layers. This enhancement enables the agent to exploit problem specific particular quantum correlations when learning policies, and so richer exploration of the routing solution space. The QAOA-augmented QRL framework shows quicker convergence in training and can tackle larger VRP instances that are beyond the reach of Grover's Adaptive Search (GAS) and Quantum Reinforcement Learning (QRL) approaches. Experiments on standard VRP instances demonstrate better solutions, fewer episodes to converge and good memory usage on near term quantum hardware simulators. These findings demonstrate QAOA- integrated QRL as a viable approach to scalable, high quality quantum-assisted combinatorial optimization. --- ### Unifying Deep Stochastic Processes for Image Enhancement Authors: Wojciech Kozłowski, Radosław Kuczbański, Kamil Adamczewski, Karol Szczypkowski, Maciej Zięba Published: 2026-05-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.01568v2 Deep stochastic processes have recently become a central paradigm for image enhancement, with many methods explicitly conditioning the stochastic trajectory on the degraded input. However, the relationship between these conditional processes and standard diffusion models remains unclear. In this work, we introduce a unified perspective on stochastic image enhancement by classifying recent methods into three families of continuous-time processes: unconditional diffusion models, Ornstein-Uhlenbeck (OU) processes, and diffusion bridges. We show that all of these approaches arise from a common stochastic differential equation (SDE) formulation. This framework makes explicit that seemingly disparate methods differ primarily in their drift and diffusion terms, terminal distributions, and boundary conditions, while schedulers and samplers constitute orthogonal design choices. Leveraging this unification, we conduct a controlled empirical study across multiple image enhancement tasks using identical architectures and training protocols. Our results reveal no consistently dominant method; instead, we identify and disentangle the specific design choices that most strongly influence performance. Finally, we release ItoVision, a modular PyTorch library that implements the unified framework and enables rapid prototyping and fair comparison of stochastic image enhancement methods. --- ### Unifying Deep Stochastic Processes for Image Enhancement Authors: Wojciech Kozłowski, Radosław Kuczbański, Kamil Adamczewski, Karol Szczypkowski, Maciej Zięba Published: 2026-05-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.01568v1 Deep stochastic processes have recently become a central paradigm for image enhancement, with many methods explicitly conditioning the stochastic trajectory on the degraded input. However, the relationship between these conditional processes and standard diffusion models remains unclear. In this work, we introduce a unified perspective on stochastic image enhancement by classifying recent methods into three families of continuous-time processes: unconditional diffusion models, Ornstein-Uhlenbeck (OU) processes, and diffusion bridges. We show that all of these approaches arise from a common stochastic differential equation (SDE) formulation. This framework makes explicit that seemingly disparate methods differ primarily in their drift and diffusion terms, terminal distributions, and boundary conditions, while schedulers and samplers constitute orthogonal design choices. Leveraging this unification, we conduct a controlled empirical study across multiple image enhancement tasks using identical architectures and training protocols. Our results reveal no consistently dominant method; instead, we identify and disentangle the specific design choices that most strongly influence performance. Finally, we release ItoVision, a modular PyTorch library that implements the unified framework and enables rapid prototyping and fair comparison of stochastic image enhancement methods. --- ### Feedback-Normalized Developer Memory for Reinforcement-Learning Coding Agents: A Safety-Gated MCP Architecture Authors: Mehmet Iscan Published: 2026-05-02 Categories: cs.SE, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2605.01567v1 Large language model (LLM) coding agents increasingly operate over repositories, terminals, tests, and execution traces across long software-engineering episodes. Persistent memory is useful, but static vector stores or generic retrieval-augmented generation (RAG) are insufficient for reinforcement-learning (RL) code development, where small details can alter Bellman targets, terminal masks, gradient flow, or validation claims. This paper presents RL Developer Memory, a local-first, Model Context Protocol (MCP)-native developer-memory architecture for RL coding agents. It treats memory selection as a logged contextual decision process: issue_match ranks candidates and records telemetry, issue_feedback maps raw labels to bounded rewards, and issue_record_resolution links verified resolutions to earlier retrieval events. A deterministic ranker remains deployed, while a contextual-bandit residual policy runs in shadow mode and can affect canary behavior only through conservative off-policy-evaluation (OPE) gates. RL/control memories require theory-to-code metadata and review-gated governance. The system is evaluated on a deterministic 200-case benchmark with RL algorithm bugs, hard negatives, review-gated RL/control cases, and low-risk failures. In the same-commit comparison, deterministic control and full shadow/OPE both achieve 80.0% expected-decision accuracy and 100.0% hard-negative suppression; the full configuration adds learning telemetry rather than accuracy gain. Static validation passed 11/11 checks; dynamic integration passed 10/10 cases. The evidence reports limits: active learned-policy deployment and official-client MCP interoperability are unsupported, live full-configuration latency regresses, and 40 residual non-RL failures remain. The contribution is an auditable memory-control architecture with explicit claim boundaries, not a universal coding-agent improvement claim. --- ### Multi-Agent Reasoning Improves Compute Efficiency: Pareto-Optimal Test-Time Scaling Authors: Florian Valentin Wunderlich, Lars Benedikt Kaesberg, Jan Philip Wahle, Terry Ruas, Bela Gipp Published: 2026-05-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2605.01566v1 Advances in inference methods have enabled language models to improve their predictions without additional training. These methods often prioritize raw performance over cost-effective compute usage. However, computational efficiency is key for real-world applications with resource constraints. We provide a systematic analysis of the inference scaling strategies self-consistency, self-refinement, multi-agent debate, and mixture-of-agents, to study their computational performance tradeoffs. We evaluate methods on two reasoning benchmarks (MMLU-Pro, BBH) and include extensive parameter configurations (e.g., scaling the number of parallel predictions, agents, and debate rounds) across different model sizes. Across 34 configurations and over 100 evaluations, we compute the Pareto-optimal front to select methods that achieve the best accuracy with the lowest computational budget. Notably, inference scaling improves accuracy by up to +7.1% points over chain-of-thought at the highest evaluated budgets (20x the CoT compute budget) on MMLU-Pro. With an equal computing budget, debate and mixture-of-agents outperform self-consistency by 1.3% and 2.7% points, respectively. While self-consistency saturates earlier, multi-agent gains persist, particularly on more complicated tasks. We identify a simple multi-agent design guideline: mixture-of-agents is most efficient when the number of parallel generations exceeds the number of sequential aggregations. --- ### Multi-Dataset Cross-Domain Knowledge Distillation for Unified Medical Image Segmentation, Classification, and Detection Authors: Ceausescu Ciprian-Mihai, Anghelina Ion-Marian, Alexe Dumitru-Bogdan Published: 2026-05-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.01563v1 We propose a unified cross-domain transfer learning framework that leverages knowledge from multiple heterogeneous medical imaging datasets to improve performance across segmentation, classification, and object detection tasks. Our approach employs a teacher-student paradigm in which a joint teacher model aggregates domain-invariant representations learned from diverse source datasets, while a task-specific student model is trained via multi-level knowledge distillation. Originally developed for medical image segmentation, the framework is extended to support image-level classification and object-level detection, enabling a general multi-task formulation for medical image analysis. We evaluate our method on a broad suite of datasets, including six segmentation benchmarks, BrainMetShare, ISLES, BraTS (MRI) and Lung MSD, LiTS, KiTS (CT), as well as multiple classification datasets for pulmonary disease and dementia, and detection datasets with native bounding-box annotations. Across all tasks and modalities, the proposed approach yields consistent improvements over strong dataset-specific and multi-head baselines, demonstrating enhanced robustness to distributional shifts and superior generalization. These findings highlight the potential of multi-dataset knowledge distillation as a scalable and task-agnostic approach for enhancing segmentation, classification, and object detection performance across heterogeneous medical imaging domains. --- ### Neuro-Symbolic Agents for Hallucination-Free Requirements Reuse Authors: Ahmed F. Ibrahim Published: 2026-05-02 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2605.01562v2 The Object-Oriented Method for Requirements Authoring and Management (OOMRAM) is a requirements reuse framework that relies on exact identifier matching and rigid templates, limiting its ability to adapt specifications across diverse contexts. While Large Language Models (LLMs) offer the flexibility to overcome this bottleneck, they introduce the risk of generating structurally invalid or inconsistent requirement combinations. To address this tension, we present a neuro-symbolic multi-agent system that re-conceptualizes requirements reuse as a Model-Driven Elicitation process. In this paradigm, an LLM serves as a non-deterministic heuristic for traversing a deterministic domain model represented by a formal OOMRAM requirement lattice. A deterministic, symbolic validator enforces all structural constraints within the agent loop, effectively eliminating hallucinated requirement combinations by construction. Evaluated on an autonomous benchmark across two application families, our system achieves 100% requirement coverage and a constraint-violation rate of only 0.2%. Although the F1-score against a single gold standard is moderate (0.47-0.51), every generated specification is structurally valid and satisfies all mandatory domain requirements. The model-agnostic implementation scales to larger lattices via subgraph navigation and provides transparent audit trails for regulatory compliance. --- ### Hall-Like Transversal Stress and Sandpile Criticality on Real Production Networks Authors: Diego Vallarino Published: 2026-05-02 Categories: econ.EM, cs.LG, physics.soc-ph arXiv: https://arxiv.org/abs/2605.01561v1 This paper develops a Hall-Sandpile model of economic instability that combines a Hall-like transversal stress mechanism with sandpile threshold dynamics on a real production-network substrate. In analogy with the physical Hall effect, where exposed flows under an external field generate stress in a transversal direction, we model economic shocks as fields that act on flow-intensive, low-redundancy, low-capacity nodes and produce systemic stress through a multiplicative conversion function. The accumulated stress drives a discrete toppling rule and an avalanche dynamics whose effective activation threshold declines with transversal exposure. The model is calibrated on annual World Input--Output Database (WIOD) production networks for 2000--2014 and simulated on the 2014 substrate (2{,}283 country--sector nodes) under three alternative propagation normalisations to avoid mechanical near-criticality from row-stochastic operators. Controlled Monte Carlo experiments over external field intensity and redundancy stress generate four ordered regimes: stable absorption, latent fragility, critical transition, and avalanche regime. Mean avalanche size and the probabilities of finite-size systemic events $\Pr(S\!\geq\!5)$, $\Pr(S\!\geq\!10)$ and $\Pr(S\!\geq\!20)$ rise jointly with field intensity and redundancy stress. Tail diagnostics show regime-dependent thickening of the avalanche distribution, but the estimated tail indices remain too high to interpret as evidence of universal power-law criticality. The contribution is therefore a finite-size, real-network description of how transversal stress activates structural fragility, not a claim of self-organised criticality in the global economy. --- ### Automated Interpretability and Feature Discovery in Language Models with Agents Authors: Arnau Marin-Llobet, Javier Ferrando Published: 2026-05-02 Categories: cs.CL, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2605.01555v1 We introduce an autonomous multiagent framework for mechanistic interpretability that automates both explaining and finding internal features in large language models. The system runs two coupled loops: (1) explanation refinement, where an agent proposes competing hypotheses and iteratively tests them with targeted prompt controls and a multi-metric evaluation; and (2) feature discovery, where an agent generates prompt sets, constructs a k-nearest-neighbor graph in activation space, and retrieves candidate features using statistical separability and semantic coherence criteria. On Gemma-2 family models and MLP neurons in weight-sparse transformers, our agent improves over one-shot auto-interpretations, discovers language-specific and safety-relevant features, and produces auditable explanation traces, showing that agent-driven empirical loops yield sharper and more falsifiable explanations than one-shot labels. --- ### GeoBlur: Epipolar Geometry Estimation from a Single Motion-Blurred Image Authors: Bao-Long Tran, Cuong Le, Fredrik Viksten, Per-Erik Forssén Published: 2026-05-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.01552v2 Relative camera pose geometry, formulated via fundamental matrix estimation, is a challenging problem in many robotics and VR/AR applications. These applications occasionally contain fast monocular camera motion, which severely blurs the image and prevents the use of traditional multi-view geometry methods for camera pose estimation. To handle these cases, we propose GeoBlur, a framework for estimating the fundamental matrix and recovering relative camera pose directly from a single motion-blurred image, using the motion cues from blur artifacts. GeoBlur first predicts the visual correspondences between two time instances within the camera exposure window; then, it infers the fundamental matrix by solving the single-frame epipolar geometry problem under time-direction ambiguity. The resulting fundamental matrix is unique up to transposition, reflecting the inherent ambiguity in the direction of time. GeoBlur improves performance on synthetic and hybrid benchmarks while remaining competitive with prior work on real motion-blur data. We further demonstrate the use of GeoBlur on the downstream task on single frame motion segmentation. --- ### Robust Fundamental Matrix Estimation from Single Image Motion Blur Authors: Bao-Long Tran, Per-Erik Forssén, Fredrik Viksten Published: 2026-05-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2605.01552v1 In this paper, we introduce a challenging task: extracting a fundamental matrix from a single motion blurred image. For a camera moving in 3D during exposure, the smear paths in the blurry image contain cues and constraints on this motion. We demonstrate the feasibility of establishing correspondences between two time instances within the camera exposure window, and that these can be used to robustly infer a fundamental matrix, which summarizes the motion of the camera during the exposure time. The inferred fundamental matrix is unique up to a transpose, corresponding to an ambiguity of the direction of time. Due to this per-smear ambiguity, classic methods, such as the 8-point algorithm, are no longer usable. The proposed method modifies the estimation to work on time-direction ambiguous correspondences. To improve the robustness of the fundamental matrix estimation, we also propose to incorporate an uncertainty measurement in smear pattern prediction and use it in the sampling process of the estimator. Experiments on synthetic and real-world motion-blur datasets demonstrate that our approach is able to estimate the fundamental matrix encoding the 3D camera motion, from single frames. Practical applicability is demonstrated on the downstream task of motion segmentation. ---