# Research Papers — 2026-04 · bundle page 143 of 233 > Full-text markdown of papers #7101–7150 of 11612 published in 2026-04, newest first. AI-readable mirrors of the full 11612-paper 2026-04 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-04/page/143 [← Previous bundle](https://aipod.dev/papers/archive/2026-04/page/142.md) [Next bundle →](https://aipod.dev/papers/archive/2026-04/page/144.md) ### Generating Multiple-Choice Knowledge Questions with Interpretable Difficulty Estimation using Knowledge Graphs and Large Language Models Authors: Mehmet Can Şakiroğlu, H. Altay Güvenir, Kamer Kaya Published: 2026-04-12 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2604.10748v1 Generating multiple-choice questions (MCQs) with difficulty estimation remains challenging in automated MCQ-generation systems used in adaptive, AI-assisted education. This study proposes a novel methodology for generating MCQs with difficulty estimation from the input documents by utilizing knowledge graphs (KGs) and large language models (LLMs). Our approach uses an LLM to construct a KG from input documents, from which MCQs are then systematically generated. Each MCQ is generated by selecting a node from the KG as the key, sampling a related triple or quintuple -- optionally augmented with an extra triple -- and prompting an LLM to generate a corresponding stem from these graph components. Distractors are then selected from the KG. For each MCQ, nine difficulty signals are computed and combined into a unified difficulty score using a data-driven approach. Experimental results demonstrate that our method generates high-quality MCQs whose difficulty estimation is interpretable and aligns with human perceptions. Our approach improves automated MCQ generation by integrating structured knowledge representations with LLMs and a data-driven difficulty estimation model. --- ### How You Ask Matters! Adaptive RAG Robustness to Query Variations Authors: Yunah Jang, Megha Sundriyal, Kyomin Jung, Meeyoung Cha Published: 2026-04-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.10745v1 Adaptive Retrieval-Augmented Generation (RAG) promises accuracy and efficiency by dynamically triggering retrieval only when needed and is widely used in practice. However, real-world queries vary in surface form even with the same intent, and their impact on Adaptive RAG remains under-explored. We introduce the first large-scale benchmark of diverse yet semantically identical query variations, combining human-written and model-generated rewrites. Our benchmark facilitates a systematic evaluation of Adaptive RAG robustness by examining its key components across three dimensions: answer quality, computational cost, and retrieval decisions. We discover a critical robustness gap, where small surface-level changes in queries dramatically alter retrieval behavior and accuracy. Although larger models show better performance, robustness does not improve accordingly. These findings reveal that Adaptive RAG methods are highly vulnerable to query variations that preserve identical semantics, exposing a critical robustness challenge. --- ### Deep-Reporter: Deep Research for Grounded Multimodal Long-Form Generation Authors: Fangda Ye, Zhifei Xie, Yuxin Hu, Yihang Yin, Shurui Huang, Shikai Dong, Jianzhu Bao, Shuicheng Yan Published: 2026-04-12 Categories: cs.CL, cs.AI, cs.IR arXiv: https://arxiv.org/abs/2604.10741v2 Recent agentic search frameworks enable deep research via iterative planning and retrieval, reducing hallucinations and enhancing factual grounding. However, they remain text-centric, overlooking the multimodal evidence that characterizes real-world expert reports. We introduce a pressing task: multimodal long-form generation. Accordingly, we propose Deep-Reporter, a unified agentic framework for grounded multimodal long-form generation. It orchestrates: (i) Agentic Multimodal Search and Filtering to retrieve and filter textual passages and information-dense visuals; (ii) Checklist-Guided Incremental Synthesis to ensure coherent image-text integration and optimal citation placement; and (iii) Recurrent Context Management to balance long-range coherence with local fluency. We develop a rigorous curation pipeline producing 8K high-quality agentic traces for model optimization. We further introduce M2LongBench, a comprehensive testbed comprising 247 research tasks across 9 domains and a stable multimodal sandbox. Extensive experiments demonstrate that long-form multimodal generation is a challenging task, especially in multimodal selection and integration, and effective post-training can bridge the gap. --- ### RCBSF: A Multi-Agent Framework for Automated Contract Revision via Stackelberg Game Authors: Shijia Xu, Yu Wang, Xiaolong Jia, Zhou Wu, Kai Liu, April Xiaowen Dong Published: 2026-04-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.10740v1 Despite the widespread adoption of Large Language Models (LLMs) in Legal AI, their utility for automated contract revision remains impeded by hallucinated safety and a lack of rigorous behavioral constraints. To address these limitations, we propose the Risk-Constrained Bilevel Stackelberg Framework (RCBSF), which formulates revision as a non-cooperative Stackelberg game. RCBSF establishes a hierarchical Leader Follower structure where a Global Prescriptive Agent (GPA) imposes risk budgets upon a follower system constituted by a Constrained Revision Agent (CRA) and a Local Verification Agent (LVA) to iteratively optimize output. We provide theoretical guarantees that this bilevel formulation converges to an equilibrium yielding strictly superior utility over unguided configurations. Empirical validation on a unified benchmark demonstrates that RCBSF achieves state-of-the-art performance, surpassing iterative baselines with an average Risk Resolution Rate (RRR) of 84.21\% while enhancing token efficiency. Our code is available at https://github.com/xjiacs/RCBSF . --- ### When More Thinking Hurts: Overthinking in LLM Test-Time Compute Scaling Authors: Shu Zhou, Rui Ling, Junan Chen, Xin Wang, Tao Fan, Hao Wang Published: 2026-04-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.10739v1 Scaling test-time compute through extended chains of thought has become a dominant paradigm for improving large language model reasoning. However, existing research implicitly assumes that longer thinking always yields better results. This assumption remains largely unexamined. We systematically investigate how the marginal utility of additional reasoning tokens changes as compute budgets increase. We find that marginal returns diminish substantially at higher budgets and that models exhibit ``overthinking'', where extended reasoning is associated with abandoning previously correct answers. Furthermore, we show that optimal thinking length varies across problem difficulty, suggesting that uniform compute allocation is suboptimal. Our cost-aware evaluation framework reveals that stopping at moderate budgets can reduce computation significantly while maintaining comparable accuracy. --- ### BlasBench: An Open Benchmark for Irish Speech Recognition Authors: Jyoutir Raj, John Conway Published: 2026-04-12 Categories: cs.CL, cs.SD arXiv: https://arxiv.org/abs/2604.10736v2 Existing multilingual benchmarks include Irish among dozens of languages but apply no Irish-aware text normalisation, leaving reliable and reproducible ASR comparison impossible. We introduce BlasBench, an open evaluation harness that provides a standalone Irish-aware normaliser preserving fadas, lenition, and eclipsis; a reproducible scoring harness and per-utterance predictions released for all evaluated runs. We pilot this by benchmarking 12 systems across four architecture families on Common Voice ga-IE and FLEURS ga-IE. All Whisper variants exceed 100% WER through insertion-driven hallucination. Microsoft Azure reaches 22.2% WER on Common Voice and 57.5% on FLEURS; the best open model, Omnilingual ASR 7B, reaches 30.65% and 39.09% respectively. Models fine-tuned on Common Voice degrade 33-43 points moving to FLEURS, while massively multilingual models degrade only 7-10 - a generalisation gap that single-dataset evaluation misses. --- ### Self-Correcting RAG: Enhancing Faithfulness via MMKP Context Selection and NLI-Guided MCTS Authors: Shijia Xu, Zhou Wu, Xiaolong Jia, Yu Wang, Kai Liu, April Xiaowen Dong Published: 2026-04-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.10734v1 Retrieval-augmented generation (RAG) substantially extends the knowledge boundary of large language models. However, it still faces two major challenges when handling complex reasoning tasks: low context utilization and frequent hallucinations. To address these issues, we propose Self-Correcting RAG, a unified framework that reformulates retrieval and generation as constrained optimization and path planning. On the input side, we move beyond traditional greedy retrieval and, for the first time, formalize context selection as a multi-dimensional multiple-choice knapsack problem (MMKP), thereby maximizing information density and removing redundancy under a strict token budget. On the output side, we introduce a natural language inference (NLI)-guided Monte Carlo Tree Search (MCTS) mechanism, which leverages test-time compute to dynamically explore reasoning trajectories and validate the faithfulness of generated answers. Experiments on six multi-hop question answering and fact-checking datasets demonstrate that our method significantly improves reasoning accuracy on complex queries while effectively reducing hallucinations, outperforming strong existing baselines.Our code is available at https://github.com/xjiacs/Self-Correcting-RAG . --- ### Too Nice to Tell the Truth: Quantifying Agreeableness-Driven Sycophancy in Role-Playing Language Models Authors: Arya Shah, Deepali Mishra, Chaklam Silpasuwanchai Published: 2026-04-12 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2604.10733v1 Large language models increasingly serve as conversational agents that adopt personas and role-play characters at user request. This capability, while valuable, raises concerns about sycophancy: the tendency to provide responses that validate users rather than prioritize factual accuracy. While prior work has established that sycophancy poses risks to AI safety and alignment, the relationship between specific personality traits of adopted personas and the degree of sycophantic behavior remains unexplored. We present a systematic investigation of how persona agreeableness influences sycophancy across 13 small, open-weight language models ranging from 0.6B to 20B parameters. We develop a benchmark comprising 275 personas evaluated on NEO-IPIP agreeableness subscales and expose each persona to 4,950 sycophancy-eliciting prompts spanning 33 topic categories. Our analysis reveals that 9 of 13 models exhibit statistically significant positive correlations between persona agreeableness and sycophancy rates, with Pearson correlations reaching $r = 0.87$ and effect sizes as large as Cohen's $d = 2.33$. These findings demonstrate that agreeableness functions as a reliable predictor of persona-induced sycophancy, with direct implications for the deployment of role-playing AI systems and the development of alignment strategies that account for personality-mediated deceptive behaviors. --- ### Perceived Importance of Cognitive Skills Among Computing Students in the Era of AI Authors: Neha Rani, Erta Cenko, Laura Melissa Cruz Castro Published: 2026-04-12 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2604.10730v1 The availability and increasing integration of generative AI tools have transformed computing education. While AI in education presents opportunities, it also raises new concerns about how these powerful know-it-all AI tools, which are becoming widespread, impact cognitive skill development among students. Cognitive skills are essential for academic success and professional competence. It relates to the ability to understand, analyze, evaluate, synthesize information and more. The extensive use of these AI tools can aid in cognitive offloading, freeing up cognitive resources to be used in other tasks and activities. However, cognitive offloading may inadvertently lead to diminishing cognitive involvement in learning and related activities when using AI tools. Understanding cognitive skills' impact in the era of AI is essential to align curricular design with evolving workforce demands and changing work environment and processes. To address this concern and to develop an understanding of how the importance of cognitive skills changes with increasing integration of AI, we conducted a researcher-monitored and regulated quantitative survey of undergraduate computing students. We examined students' perceptions of cognitive skills across three temporal frames: prior to widespread AI adoption (past), current informal and formal use of AI in learning contexts (present), and future with even more AI integration in professional environments (future). In the study, students rated the importance of 11 cognitive skills. Our analysis reveals that students expect all 11 cognitive skills to be of diminishing importance in the future, when AI use and integration increases. Our findings highlight the need for educational interventions that explicitly reinforce cognitive skill development within learning environments that are now often relying on AI. --- ### Tail-Aware Information-Theoretic Bounds for LLM Alignment under Heavy-Tailed Rewards Authors: Huiming Zhang, Binghan Li, Wan Tian, Qiang Sun Published: 2026-04-12 Categories: stat.ML, cs.AI, cs.LG, math.PR, math.ST arXiv: https://arxiv.org/abs/2604.10727v2 Classical information-theoretic learning bounds typically rely on KL mutual information and moment-generating-function (MGF) arguments, which are well matched to bounded or sub-Gaussian losses but can be ineffective when losses or rewards are heavy-tailed. We develop a tail-aware information-theoretic framework for sub-Weibull data, where the tail parameter $θ$ controls the tail heaviness: $θ=2$ corresponds to sub-Gaussian, $θ=1$ to sub-exponential, and $0<θ<1$ to genuinely heavy tails. Our key technical ingredient is a decorrelation lemma that bounds change-of-measure expectations using a shifted-log $f_θ$-divergence, which admits explicit comparisons to Rényi divergence without MGF arguments. On the empirical-process side, we establish sharp maximal inequalities and a Dudley-type chaining bound for sub-Weibull processes, with logarithmic and entropy terms raised to the power $1/θ$. These tools yield tail-adaptive selection bounds and a multiscale information-theoretic Dudley inequality based on shifted-log and Rényi mutual information. We apply our theory to large language models (LLMs) in the context of reward hacking within reinforcement learning from human feedback (RLHF). We show that Rényi-regularized alignment provides finite reward guarantees and ensures that best-of-N policies remain well-controlled, thereby mitigating the catastrophic Goodhart effects where standard KL-regularization fails. We illustrate Rényi-regularized RLHF by experiments, including controlled heavy-tailed rewards and token-space reward attacks. --- ### Expect the Unexpected? Testing the Surprisal of Salient Entities Authors: Jessica Lin, Amir Zeldes Published: 2026-04-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.10724v1 Previous work examining the Uniform Information Density (UID) hypothesis has shown that while information as measured by surprisal metrics is distributed more or less evenly across documents overall, local discrepancies can arise due to functional pressures corresponding to syntactic and discourse structural constraints. However, work thus far has largely disregarded the relative salience of discourse participants. We fill this gap by studying how overall salience of entities in discourse relates to surprisal using 70K manually annotated mentions across 16 genres of English and a novel minimal-pair prompting method. Our results show that globally salient entities exhibit significantly higher surprisal than non-salient ones, even controlling for position, length, and nesting confounds. Moreover, salient entities systematically reduce surprisal for surrounding content when used as prompts, enhancing document-level predictability. This effect varies by genre, appearing strongest in topic-coherent texts and weakest in conversational contexts. Our findings refine the UID competing pressures framework by identifying global entity salience as a mechanism shaping information distribution in discourse. --- ### Turning Generators into Retrievers: Unlocking MLLMs for Natural Language-Guided Geo-Localization Authors: Yuqi Chen, Xiaohan Zhang, Ahmad Arrabi, Waqas Sultani, Chen Chen, Safwan Wshah Published: 2026-04-12 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2604.10721v1 Natural-language Guided Cross-view Geo-localization (NGCG) aims to retrieve geo-tagged satellite imagery using textual descriptions of ground scenes. While recent NGCG methods commonly rely on CLIP-style dual-encoder architectures, they often suffer from weak cross-modal generalization and require complex architectural designs. In contrast, Multimodal Large Language Models (MLLMs) offer powerful semantic reasoning capabilities but are not directly optimized for retrieval tasks. In this work, we present a simple yet effective framework to adapt MLLMs for NGCG via parameter-efficient finetuning. Our approach optimizes latent representations within the MLLM while preserving its pretrained multimodal knowledge, enabling strong cross-modal alignment without redesigning model architectures. Through systematic analysis of diverse variables, from model backbone to feature aggregation, we provide practical and generalizable insights for leveraging MLLMs in NGCG. Our method achieves SOTA on GeoText-1652 with a 12.2% improvement in Text-to-Image Recall@1 and secures top performance in 5 out of 12 subtasks on CVG-Text, all while surpassing baselines with far fewer trainable parameters. These results position MLLMs as a robust foundation for semantic cross-view retrieval and pave the way for MLLM-based NGCG to be adopted as a scalable, powerful alternative to traditional dual-encoder designs. Project page and code are available at https://yuqichen888.github.io/NGCG-MLLMs-web/. --- ### Teaching Language Models How to Code Like Learners: Conversational Serialization for Student Simulation Authors: Charles Koutcheme, Juho Leinonen, Arto Hellas Published: 2026-04-12 Categories: cs.AI, cs.CL, cs.CY arXiv: https://arxiv.org/abs/2604.10720v2 Artificial students -- models that simulate how learners act and respond within educational systems -- are a promising tool for evaluating tutoring strategies and feedback mechanisms at scale. However, most existing approaches rely on prompting large, proprietary language models, limiting adaptability to specific courses and raising concerns around privacy, cost, and dependence. In this work, we propose a framework for training open-weight artificial programming learners directly from authentic student process data. Our approach serializes temporal log traces into a conversational format, representing each student's problem-solving process as a dialogue between the learner and their automated assessment system. Student code submissions and environment feedback, such as test outcomes, grades, and error traces, form alternating conversational turns, enabling models to learn from the iterative debugging process. We additionally introduce a training pipeline combining supervised fine-tuning with preference optimization to align models with authentic student debugging behavior. We evaluate our framework by training Qwen models at 4B and 8B scales on a large-scale dataset of real student submissions to Python programming assignments. Our results show that incorporating environment feedback strengthens models' ability to replicate student debugging behavior, improving over both prior code-only approaches and prompted large language models baselines in functional alignment and code similarity. We release our code to support reproducibility. --- ### SciPredict: Can LLMs Predict the Outcomes of Scientific Experiments in Natural Sciences? Authors: Udari Madhushani Sehwag, Elaine Lau, Haniyeh Ehsani Oskouie, Shayan Shabihi, Erich Liang, Andrea Toledo, Guillermo Mangialardi, Sergio Fonrouge, Ed-Yeremai Hernandez Cardona, Paula Vergara, Utkarsh Tyagi, Chen Bo Calvin Zhang et al. Published: 2026-04-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.10718v1 Accelerating scientific discovery requires the identification of which experiments would yield the best outcomes before committing resources to costly physical validation. While existing benchmarks evaluate LLMs on scientific knowledge and reasoning, their ability to predict experimental outcomes - a task where AI could significantly exceed human capabilities - remains largely underexplored. We introduce SciPredict, a benchmark comprising 405 tasks derived from recent empirical studies in 33 specialized sub-fields of physics, biology, and chemistry. SciPredict addresses two critical questions: (a) can LLMs predict the outcome of scientific experiments with sufficient accuracy? and (b) can such predictions be reliably used in the scientific research process? Evaluations reveal fundamental limitations on both fronts. Model accuracies are 14-26% and human expert performance is $\approx$20%. Although some frontier models exceed human performance model accuracy is still far below what would enable reliable experimental guidance. Even within the limited performance, models fail to distinguish reliable predictions from unreliable ones, achieving only $\approx$20% accuracy regardless of their confidence or whether they judge outcomes as predictable without physical experimentation. Human experts, in contrast, demonstrate strong calibration: their accuracy increases from $\approx$5% to $\approx$80% as they deem outcomes more predictable without conducting the experiment. SciPredict establishes a rigorous framework demonstrating that superhuman performance in experimental science requires not just better predictions, but better awareness of prediction reliability. For reproducibility all our data and code are provided at https://github.com/scaleapi/scipredict --- ### Detecting RAG Extraction Attack via Dual-Path Runtime Integrity Game Authors: Yuanbo Xie, Yingjie Zhang, Yulin Li, Shouyou Song, Xiaokun Chen, Zhihan Liu, Liya Su, Tingwen Liu Published: 2026-04-12 Categories: cs.CR, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2604.10717v1 Retrieval-Augmented Generation (RAG) systems augment large language models with external knowledge, yet introduce a critical security vulnerability: RAG Knowledge Base Leakage, wherein adversarial prompts can induce the model to divulge retrieved proprietary content. Recent studies reveal that such leakage can be executed through adaptive and iterative attack strategies (named RAG extraction attack), while effective countermeasures remain notably lacking. To bridge this gap, we propose CanaryRAG, a runtime defense mechanism inspired by stack canaries in software security. CanaryRAG embeds carefully designed canary tokens into retrieved chunks and reformulates RAG extraction defense as a dual-path runtime integrity game. Leakage is detected in real time whenever either the target or oracle path violates its expected canary behavior, including under adaptive suppression and obfuscation. Extensive evaluations against existing attacks demonstrate that CanaryRAG provides robust defense, achieving substantially lower chunk recovery rates than state-of-the-art baselines while imposing negligible impact on task performance and inference latency. Moreover, as a plug-and-play solution, CanaryRAG can be seamlessly integrated into arbitrary RAG pipelines without requiring retraining or structural modifications, offering a practical and scalable safeguard for proprietary data. --- ### Defending against Patch-Based and Texture-Based Adversarial Attacks with Spectral Decomposition Authors: Wei Zhang, Xinyu Chang, Xiao Li, Yiming Zhu, Xiaolin Hu Published: 2026-04-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.10715v1 Adversarial examples present significant challenges to the security of Deep Neural Network (DNN) applications. Specifically, there are patch-based and texture-based attacks that are usually used to craft physical-world adversarial examples, posing real threats to security-critical applications such as person detection in surveillance and autonomous systems, because those attacks are physically realizable. Existing defense mechanisms face challenges in the adaptive attack setting, i.e., the attacks are specifically designed against them. In this paper, we propose Adversarial Spectrum Defense (ASD), a defense mechanism that leverages spectral decomposition via Discrete Wavelet Transform (DWT) to analyze adversarial patterns across multiple frequency scales. The multi-resolution and localization capability of DWT enables ASD to capture both high-frequency (fine-grained) and low-frequency (spatially pervasive) perturbations. By integrating this spectral analysis with the off-the-shelf Adversarial Training (AT) model, ASD provides a comprehensive defense strategy against both patch-based and texture-based adversarial attacks. Extensive experiments demonstrate that ASD+AT achieved state-of-the-art (SOTA) performance against various attacks, outperforming the APs of previous defense methods by 21.73%, in the face of strong adaptive adversaries specifically designed against ASD. Code available at https://github.com/weiz0823/adv-spectral-defense . --- ### Audio-Omni: Extending Multi-modal Understanding to Versatile Audio Generation and Editing Authors: Zeyue Tian, Binxin Yang, Zhaoyang Liu, Jiexuan Zhang, Ruibin Yuan, Hubery Yin, Qifeng Chen, Chen Li, Jing Lyu, Wei Xue, Yike Guo Published: 2026-04-12 Categories: cs.SD, cs.AI, cs.CV, cs.MM arXiv: https://arxiv.org/abs/2604.10708v2 Recent progress in multimodal models has spurred rapid advances in audio understanding, generation, and editing. However, these capabilities are typically addressed by specialized models, leaving the development of a truly unified framework that can seamlessly integrate all three tasks underexplored. While some pioneering works have explored unifying audio understanding and generation, they often remain confined to specific domains. To address this, we introduce Audio-Omni, the first end-to-end framework to unify generation and editing across general sound, music, and speech domains, with integrated multi-modal understanding capabilities. Our architecture synergizes a frozen Multimodal Large Language Model for high-level reasoning with a trainable Diffusion Transformer for high-fidelity synthesis. To overcome the critical data scarcity in audio editing, we construct AudioEdit, a new large-scale dataset comprising over one million meticulously curated editing pairs. Extensive experiments demonstrate that Audio-Omni achieves state-of-the-art performance across a suite of benchmarks, outperforming prior unified approaches while achieving performance on par with or superior to specialized expert models. Beyond its core capabilities, Audio-Omni exhibits remarkable inherited capabilities, including knowledge-augmented reasoning generation, in-context generation, and zero-shot cross-lingual control for audio generation, highlighting a promising direction toward universal generative audio intelligence. The code, model, and dataset will be publicly released on https://zeyuet.github.io/Audio-Omni. --- ### Investigating Bias and Fairness in Appearance-based Gaze Estimation Authors: Burak Akgül, Erol Şahin, Sinan Kalkan Published: 2026-04-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.10707v1 While appearance-based gaze estimation has achieved significant improvements in accuracy and domain adaptation, the fairness of these systems across different demographic groups remains largely unexplored. To date, there is no comprehensive benchmark quantifying algorithmic bias in gaze estimation. This paper presents the first extensive evaluation of fairness in appearance-based gaze estimation, focusing on ethnicity and gender attributes. We establish a fairness baseline by analyzing state-of-the-art models using standard fairness metrics, revealing significant performance disparities. Furthermore, we evaluate the effectiveness of existing bias mitigation strategies when applied to the gaze domain and show that their fairness contributions are limited. We summarize key insights and open issues. Overall, our work calls for research into developing robust, equitable gaze estimators. To support future research and reproducibility, we publicly release our annotations, code, and trained models at: github.com/akgulburak/gaze-estimation-fairness --- ### INCRT: An Incremental Transformer That Determines Its Own Architecture Authors: Giansalvo Cirrincione Published: 2026-04-12 Categories: cs.LG, cs.NE arXiv: https://arxiv.org/abs/2604.10703v1 Transformer architectures are designed by trial and error: the number of attention heads, the depth, and the head size are fixed before training begins, with no mathematical principle to guide the choice. The result is systematic structural redundancy -- between half and four-fifths of all heads in a trained model can be removed without measurable loss -- because the architecture allocates capacity without reference to the actual requirements of the task.This paper introduces INCRT (Incremental Transformer), an architecture that determines its own structure during training. Starting from a single head, INCRT adds one attention head at a time whenever its current configuration is provably insufficient, and prunes heads that have become redundant. Each growth decision is driven by a single, online-computable geometric quantity derived from the task's directional structure, requiring no separate validation phase and no hand-tuned schedule. Two theorems form the theoretical backbone. The first (homeostatic convergence) establishes that the system always reaches a finite stopping configuration that is simultaneously minimal (no redundant heads) and sufficient (no uncaptured directional energy above the threshold). The second (compressed-sensing analogy) provides a geometric upper bound on the number of heads that this configuration can contain, as a function of the spectral complexity of the task. Experiments on SARS-CoV-2 variant classification and SST-2 sentiment analysis confirm both results: the predicted and observed head counts agree within 12% across all benchmarks, and the final architectures match or exceed BERT-base on distribution-specific tasks while using between three and seven times fewer parameters and no pre-training. --- ### Backbone-Conditional Behavior of Modality Gating in Multi-Modal Prostate MRI Segmentation: A 5-Fold Cross-Validation and Gate Mechanism Analysis Authors: Yongbo Shu, Wenzhao Xie, Shanhu Yao, Zirui Xin, Luo Lei, Kewen Chen, Aijing Luo Published: 2026-04-12 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2604.10702v4 Robust segmentation of clinically significant prostate cancer (csPCa) on multi-parametric MRI must tolerate frequent degradation of its most informative diffusion sequences. Multi-modal fusion commonly employs learned modality gating under the assumption that gates implement per-sample modality quality routing -- rarely tested directly. We ask how gating behaves across backbone architectures. We systematically analyze modality-isolated gated fusion (MIGF) for csPCa segmentation on two backbones (nnU-Net and Mamba) using PI-CAI (n=1500), with cross-cohort validation on Prostate158 (n=158): a factorial ablation over gating, modality dropout, and deep supervision under 5-fold cross-validation (180 trained models), plus a gate-weight and counterfactual analysis of 30 trained gating models. Modality gating is backbone-conditional. On nnU-Net, adding gating reduces the ranking score (marginal effect -0.037; gating configurations p<0.05), whereas on Mamba the gating-plus-dropout configuration improves it (+0.024, p=0.037). Gate-weight analysis explains this: nnU-Net gates collapse into a near-static modality prior (across-case SD 0.0033), while Mamba gates retain sample-dependent variation (0.0365, ~11x larger, non-overlapping); replacing per-sample gates with their training-set mean leaves nnU-Net unchanged but degrades Mamba. Modality dropout is the only component beneficial on both backbones. Under cross-cohort shift, convolutional backbones collapse to case-level specificity near zero, whereas Mamba retains it (MIGF-Mamba highest, 0.31). Learned modality gates do not universally perform per-sample quality routing; their effective behavior is conditional on the backbone's inherent modality awareness. Among tested configurations, MIGF-Mamba is the most cross-cohort robust, and training-time modality dropout is the only component beneficial across both backbones. --- ### Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning Authors: Zikang Shan, Han Zhong, Liwei Wang, Li Zhao Published: 2026-04-12 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2604.10701v2 Credit assignment is a central challenge in reinforcement learning (RL). Classical actor-critic methods address this challenge through fine-grained advantage estimation based on a learned value function. However, learned value models are often avoided in modern large language model (LLM) RL because conventional discriminative critics are difficult to train reliably. We revisit value modeling and argue that this difficulty is partly due to limited expressiveness. In particular, representation complexity theory suggests that value functions can be hard to approximate under the one-shot prediction paradigm used by existing value models, and our scaling experiments show that such critics do not improve reliably with scale. Motivated by this observation, we propose Generative Actor-Critic (GenAC), which replaces one-shot scalar value prediction with a generative critic that performs chain-of-thought reasoning before producing a value estimate. We further introduce In-Context Conditioning, which helps the critic remain calibrated to the current actor throughout training. GenAC improves value approximation, ranking reliability, and out-of-distribution generalization, and these gains translate into stronger downstream RL performance than both value-based and value-free baselines. Overall, our results suggest that stronger value modeling is a promising direction for improving credit assignment in LLM reinforcement learning. --- ### Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning Authors: Zikang Shan, Han Zhong, Liwei Wang, Li Zhao Published: 2026-04-12 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2604.10701v1 Credit assignment is a central challenge in reinforcement learning (RL). Classical actor-critic methods address this challenge through fine-grained advantage estimation based on a learned value function. However, learned value models are often avoided in modern large language model (LLM) RL because conventional discriminative critics are difficult to train reliably. We revisit value modeling and argue that this difficulty is partly due to limited expressiveness. In particular, representation complexity theory suggests that value functions can be hard to approximate under the one-shot prediction paradigm used by existing value models, and our scaling experiments show that such critics do not improve reliably with scale. Motivated by this observation, we propose Generative Actor-Critic (GenAC), which replaces one-shot scalar value prediction with a generative critic that performs chain-of-thought reasoning before producing a value estimate. We further introduce In-Context Conditioning, which helps the critic remain calibrated to the current actor throughout training. GenAC improves value approximation, ranking reliability, and out-of-distribution generalization, and these gains translate into stronger downstream RL performance than both value-based and value-free baselines. Overall, our results suggest that stronger value modeling is a promising direction for improving credit assignment in LLM reinforcement learning. --- ### Attention Sinks as Internal Signals for Hallucination Detection in Large Language Models Authors: Jakub Binkowski, Kamil Adamczewski, Tomasz Kajdanowicz Published: 2026-04-12 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2604.10697v2 Large language models frequently exhibit hallucinations: fluent and confident outputs that are factually incorrect or unsupported by the input context. While recent hallucination detection methods have explored various features derived from attention maps, the underlying mechanisms they exploit remain poorly understood. In this work, we propose SinkProbe, a hallucination detection method grounded in the observation that hallucinations are deeply entangled with attention sinks - tokens that accumulate disproportionate attention mass during generation - indicating a transition from distributed, input-grounded attention to compressed, prior-dominated computation. Importantly, although sink scores are computed solely from attention maps, we find that the classifier preferentially relies on sinks whose associated value vectors have large norms. Moreover, we show that previous methods implicitly depend on attention sinks by establishing their mathematical relationship to sink scores. Our findings yield a novel hallucination detection method grounded in theory that produces state-of-the-art results across popular datasets and LLMs. --- ### Camyla: Scaling Autonomous Research in Medical Image Segmentation Authors: Yifan Gao, Haoyue Li, Feng Yuan, Xin Gao, Weiran Huang, Xiaosong Wang Published: 2026-04-12 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2604.10696v1 We present Camyla, a system for fully autonomous research within the scientific domain of medical image segmentation. Camyla transforms raw datasets into literature-grounded research proposals, executable experiments, and complete manuscripts without human intervention. Autonomous experimentation over long horizons poses three interrelated challenges: search effort drifts toward unpromising directions, knowledge from earlier trials degrades as context accumulates, and recovery from failures collapses into repetitive incremental fixes. To address these challenges, the system combines three coupled mechanisms: Quality-Weighted Branch Exploration for allocating effort across competing proposals, Layered Reflective Memory for retaining and compressing cross-trial knowledge at multiple granularities, and Divergent Diagnostic Feedback for diversifying recovery after underperforming trials. The system is evaluated on CamylaBench, a contamination-free benchmark of 31 datasets constructed exclusively from 2025 publications, under a strict zero-intervention protocol across two independent runs within a total of 28 days on an 8-GPU cluster. Across the two runs, Camyla generates more than 2,700 novel model implementations and 40 complete manuscripts, and surpasses the strongest per-dataset baseline selected from 14 established architectures, including nnU-Net, on 22 and 18 of 31 datasets under identical training budgets, respectively (union: 24/31). Senior human reviewers score the generated manuscripts at the T1/T2 boundary of contemporary medical imaging journals. Relative to automated baselines, Camyla outperforms AutoML and NAS systems on aggregate segmentation performance and exceeds six open-ended research agents on both task completion and baseline-surpassing frequency. These results suggest that domain-scale autonomous research is achievable in medical image segmentation. --- ### Retrieving to Recover: Towards Incomplete Audio-Visual Question Answering via Semantic-consistent Purification Authors: Jiayu Zhang, Shuo Ye, Qilang Ye, Zihan Song, Jiajian Huang, Zitong Yu Published: 2026-04-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.10695v2 Recent Audio-Visual Question Answering (AVQA) methods have advanced significantly. However, most AVQA methods lack effective mechanisms for handling missing modalities, suffering from severe performance degradation in real-world scenarios with data interruptions. Furthermore, prevailing methods for handling missing modalities predominantly rely on generative imputation to synthesize missing features. While partially effective, these methods tend to capture inter-modal commonalities but struggle to acquire unique, modality-specific knowledge within the missing data, leading to hallucinations and compromised reasoning accuracy. To tackle these challenges, we propose R$^{2}$ScP, a novel framework that shifts the paradigm of missing modality handling from traditional generative imputation to retrieval-based recovery. Specifically, we leverage cross-modal retrieval via unified semantic embeddings to acquire missing domain-specific knowledge. To maximize semantic restoration, we introduce a context-aware adaptive purification mechanism that eliminates latent semantic noise within the retrieved data. Additionally, we employ a two-stage training strategy to explicitly model the semantic relationships between knowledge from different sources. Extensive experiments demonstrate that R$^{2}$ScP significantly improves AVQA and enhances robustness in modal-incomplete scenarios. --- ### FACT-E: Causality-Inspired Evaluation for Trustworthy Chain-of-Thought Reasoning Authors: Yuxi Sun, Aoqi Zuo, Haotian Xie, Wei Gao, Mingming Gong, Jing Ma Published: 2026-04-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.10693v2 Chain-of-Thought (CoT) prompting has improved LLM reasoning, but models often generate explanations that appear coherent while containing unfaithful intermediate steps. Existing self-evaluation approaches are prone to inherent biases: the model may confidently endorse coherence even when the step-to-step implication is not valid, leading to unreliable faithfulness evaluation. We propose FACT-E, a causality-inspired framework for evaluating CoT quality. FACT-E uses controlled perturbations as an instrumental signal to separate genuine step-to-step dependence from bias-driven artifacts, producing more reliable faithfulness estimates (\textit{intra-chain faithfulness}). To select trustworthy trajectories, FACT-E jointly considers \textit{intra-chain faithfulness} and \textit{CoT-to-answer consistency}, ensuring that selected chains are both faithful internally and supportive of the correct final answer. Experiments on GSM8K, MATH, and CommonsenseQA show that FACT-E improves reasoning-trajectory selection and yields stronger in-context learning exemplars. FACT-E also reliably detects flawed reasoning under noisy conditions, providing a robust metric for trustworthy LLM reasoning. --- ### Do LLMs Build Spatial World Models? Evidence from Grid-World Maze Tasks Authors: Weijiang Li, Yilin Zhu, Rajarshi Das, Parijat Dube Published: 2026-04-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.10690v1 Foundation models have shown remarkable performance across diverse tasks, yet their ability to construct internal spatial world models for reasoning and planning remains unclear. We systematically evaluate the spatial understanding of large language models through maze tasks, a controlled testing context requiring multi-step planning and spatial abstraction. Across comprehensive experiments with Gemini-2.5-Flash, GPT-5-mini, Claude-Haiku-4.5, and DeepSeek-Chat, we uncover significant discrepancies in spatial reasoning that challenge assumptions about LLM planning capabilities. Using chain-of-thought prompting, Gemini achieves 80-86% accuracy on smaller mazes (5x5 to 7x7 grids) with tokenized adjacency representations, but performance collapses to 16-34% with visual grid formats, which is a 2-5x difference, suggesting representation-dependent rather than format-invariant spatial reasoning. We further probe spatial understanding through sequential proximity questions and compositional distance comparisons. Despite achieving 96-99% semantic coverage in reasoning traces, models fail to leverage this understanding for consistent spatial computations, indicating that they treat each question independently rather than building cumulative spatial knowledge. Our findings based on the maze-solving tasks suggest that LLMs do not develop robust spatial world models, but rather exhibit representation-specific and prompting-dependent reasoning that succeeds only under narrow conditions. These results have critical implications for deploying foundation models in applications requiring spatial abstraction. --- ### Communication-Efficient Gluon in Federated Learning Authors: Xun Qian, Alexander Gaponov, Grigory Malinovsky, Peter Richtárik Published: 2026-04-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.10689v1 Recent developments have shown that Muon-type optimizers based on linear minimization oracles (LMOs) over non-Euclidean norm balls have the potential to get superior practical performance than Adam-type methods in the training of large language models. Since large-scale neural networks are trained across massive machines, communication cost becomes the bottleneck. To address this bottleneck, we investigate Gluon, which is an extension of Muon under the more general layer-wise $(L^0, L^1)$-smooth setting, with both unbiased and contraction compressors. In order to reduce the compression error, we employ the variance reduced technique in SARAH in our compressed methods. The convergence rates and improved communication cost are achieved under certain conditions. As a byproduct, a new variance reduced algorithm with faster convergence rate than Gluon is obtained. We also incorporate momentum variance reduction (MVR) to these compressed algorithms and comparable communication cost is derived under weaker conditions when $L_i^1 \neq 0$. Finally, several numerical experiments are conducted to verify the superior performance of our compressed algorithms in terms of communication cost. --- ### SCOPE: Signal-Calibrated On-Policy Distillation Enhancement with Dual-Path Adaptive Weighting Authors: Binbin Zheng, Xing Ma, Yiheng Liang, Jingqing Ruan, Xiaoliang Fu, Kepeng Lin, Benchang Zhu, Ke Zeng, Xunliang Cai Published: 2026-04-12 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2604.10688v2 On-policy reinforcement learning has become the dominant paradigm for reasoning alignment in large language models, yet its sparse, outcome-level rewards make token-level credit assignment notoriously difficult. On-Policy Distillation (OPD) alleviates this by introducing dense, token-level KL supervision from a teacher model, but typically applies this supervision uniformly across all rollouts, ignoring fundamental differences in signal quality. We propose Signal-Calibrated On-Policy Distillation Enhancement (SCOPE), a dual-path adaptive training framework that routes on-policy rollouts by correctness into two complementary supervision paths. For incorrect trajectories, SCOPE performs teacher-perplexity-weighted KL distillation to prioritize instances where the teacher demonstrates genuine corrective capability, while down-weighting unreliable guidance. For correct trajectories, it applies student-perplexity-weighted MLE to concentrate reinforcement on low-confidence samples at the capability boundary rather than over-reinforcing already mastered ones. Both paths employ a group-level normalization to adaptively calibrate weight distributions, accounting for the intrinsic difficulty variance across prompts. Extensive experiments on six reasoning benchmarks show that SCOPE achieves an average relative improvement of 11.42% in Avg@32 and 7.30% in Pass@32 over competitive baselines, demonstrating its consistent effectiveness. --- ### QFS-Composer: Query-focused summarization pipeline for less resourced languages Authors: Vuk Đuranović, Marko Robnik Šikonja Published: 2026-04-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.10687v1 Large language models (LLMs) demonstrate strong performance in text summarization, yet their effectiveness drops significantly across languages with restricted training resources. This work addresses the challenge of query-focused summarization (QFS) in less-resourced languages, where labeled datasets and evaluation tools are limited. We present a novel QFS framework, QFS-Composer, that integrates query decomposition, question generation (QG), question answering (QA), and abstractive summarization to improve the factual alignment of a summary with user intent. We test our approach on the Slovenian language. To enable high-quality supervision and evaluation, we develop the Slovenian QA and QG models based on a Slovene LLM and adapt evaluation approaches for reference-free summary evaluation. Empirical evaluation shows that the QA-guided summarization pipeline yields improved consistency and relevance over baseline LLMs. Our work establishes an extensible methodology for advancing QFS in less-resourced languages. --- ### Critical-CoT: A Robust Defense Framework against Reasoning-Level Backdoor Attacks in Large Language Models Authors: Vu Tuan Truong, Long Bao Le Published: 2026-04-12 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2604.10681v2 Large Language Models (LLMs), despite their impressive capabilities across domains, have been shown to be vulnerable to backdoor attacks. Prior backdoor strategies predominantly operate at the token level, where an injected trigger causes the model to generate a specific target word, choice, or class (depending on the task). Recent advances, however, exploit the long-form reasoning tendencies of modern LLMs to conduct reasoning-level backdoors: once triggered, the victim model inserts one or more malicious reasoning steps into its chain-of-thought (CoT). These attacks are substantially harder to detect, as the backdoored answer remains plausible and consistent with the poisoned reasoning trajectory. Yet, defenses tailored to this type of backdoor remain largely unexplored. To bridge this gap, we propose Critical-CoT, a novel defense mechanism that conducts a two-stage fine-tuning (FT) process on LLMs to develop critical thinking behaviors, enabling them to automatically identify potential backdoors and refuse to generate malicious reasoning steps. Extensive experiments across multiple LLMs and datasets demonstrate that Critical-CoT provides strong robustness against both in-context learning-based and FT-based backdoor attacks. Notably, Critical-CoT exhibits strong cross-domain and cross-task generalization. Our code is available at hthttps://github.com/tuanvu171/Critical-CoT. --- ### FedRio: Personalized Federated Social Bot Detection via Cooperative Reinforced Contrastive Adversarial Distillation Authors: Yingguang Yang, Hao Liu, Xin Zhang, Yunhui Liu, Yutong Xia, Qi Wu, Hao Peng, Taoran Liang, Bin Chong, Tieke He, Philip S. Yu Published: 2026-04-12 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2604.10678v1 Social bot detection is critical to the stability and security of online social platforms. However, current state-of-the-art bot detection models are largely developed in isolation, overlooking the benefits of leveraging shared detection patterns across platforms to improve performance and promptly identify emerging bot variants. The heterogeneity of data distributions and model architectures further complicates the design of an effective cross-platform and cross-model detection framework. To address these challenges, we propose FedRio (Personalized Federated Social Bot Detection with Cooperative Reinforced Contrastive Adversarial Distillation framework. We first introduce an adaptive message-passing module as the graph neural network backbone for each client. To facilitate efficient knowledge sharing of global data distributions, we design a federated knowledge extraction mechanism based on generative adversarial networks. Additionally, we employ a multi-stage adversarial contrastive learning strategy to enforce feature space consistency among clients and reduce divergence between local and global models. Finally, we adopt adaptive server-side parameter aggregation and reinforcement learning-based client-side parameter control to better accommodate data heterogeneity in heterogeneous federated settings. Extensive experiments on two real-world social bot detection benchmarks demonstrate that FedRio consistently outperforms state-of-the-art federated learning baselines in detection accuracy, communication efficiency, and feature space consistency, while remaining competitive with published centralized results under substantially stronger privacy constraints. --- ### LIDEA: Human-to-Robot Imitation Learning via Implicit Feature Distillation and Explicit Geometry Alignment Authors: Yifu Xu, Bokai Lin, Xinyu Zhan, Hongjie Fang, Yong-Lu Li, Cewu Lu, Lixin Yang Published: 2026-04-12 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2604.10677v1 Scaling up robot learning is hindered by the scarcity of robotic demonstrations, whereas human videos offer a vast, untapped source of interaction data. However, bridging the embodiment gap between human hands and robot arms remains a critical challenge. Existing cross-embodiment transfer strategies typically rely on visual editing, but they often introduce visual artifacts due to intrinsic discrepancies in visual appearance and 3D geometry. To address these limitations, we introduce LIDEA (Implicit Feature Distillation and Explicit Geometric Alignment), an imitation learning framework in which policy learning benefits from human demonstrations. In the 2D visual domain, LIDEA employs a dual-stage transitive distillation pipeline that aligns human and robot representations in a shared latent space. In the 3D geometric domain, we propose an embodiment-agnostic alignment strategy that explicitly decouples embodiment from interaction geometry, ensuring consistent 3D-aware perception. Extensive experiments empirically validate LIDEA from two perspectives: data efficiency and OOD robustness. Results show that human data substitutes up to 80% of costly robot demonstrations, and the framework successfully transfers unseen patterns from human videos for out-of-distribution generalization. --- ### HiddenObjects: Scalable Diffusion-Distilled Spatial Priors for Object Placement Authors: Marco Schouten, Ioannis Siglidis, Serge Belongie, Dim P. Papadopoulos Published: 2026-04-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.10675v1 We propose a method to learn explicit, class-conditioned spatial priors for object placement in natural scenes by distilling the implicit placement knowledge encoded in text-conditioned diffusion models. Prior work relies either on manually annotated data, which is inherently limited in scale, or on inpainting-based object-removal pipelines, whose artifacts promote shortcut learning. To address these limitations, we introduce a fully automated and scalable framework that evaluates dense object placements on high-quality real backgrounds using a diffusion-based inpainting pipeline. With this pipeline, we construct HiddenObjects, a large-scale dataset comprising 27M placement annotations, evaluated across 27k distinct scenes, with ranked bounding box insertions for different images and object categories. Experimental results show that our spatial priors outperform sparse human annotations on a downstream image editing task (3.90 vs. 2.68 VLM-Judge), and significantly surpass existing placement baselines and zero-shot Vision-Language Models for object placement. Furthermore, we distill these priors into a lightweight model for fast practical inference (230,000x faster). --- ### Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents Authors: Hao Wang, Guozhi Wang, Han Xiao, Yufeng Zhou, Yue Pan, Jichao Wang, Ke Xu, Yafei Wen, Xiaohu Ruan, Xiaoxin Chen, Honggang Qi Published: 2026-04-12 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2604.10674v1 Reinforcement learning (RL) has been widely used to train LLM agents for multi-turn interactive tasks, but its sample efficiency is severely limited by sparse rewards and long horizons. On-policy self-distillation (OPSD) alleviates this by providing dense token-level supervision from a privileged teacher that has access to ground-truth answers. However, such fixed privileged information cannot capture the diverse valid strategies in agent tasks, and naively combining OPSD with RL often leads to training collapse. To address these limitations, we introduce Skill-SD, a framework that turns the agent's own trajectories into dynamic training-only supervision. Completed trajectories are summarized into compact natural language skills that describe successful behaviors, mistakes, and workflows. These skills serve as dynamic privileged information conditioning only the teacher, while the student always acts under the plain task prompt and learns to internalize the guidance through distillation. To stabilize the training, we derive an importance-weighted reverse-KL loss to provide gradient-correct token-level distillation, and dynamically synchronize the teacher with the improving student. Experimental results on agentic benchmarks demonstrate that Skill-SD substantially outperforms the standard RL baseline, improving both vanilla GRPO (+14.0%/+10.9% on AppWorld/Sokoban) and vanilla OPD (+42.1%/+40.6%). Project page: https://k1xe.github.io/skill-sd/ --- ### Principles Do Not Apply Themselves: A Hermeneutic Perspective on AI Alignment Authors: Behrooz Razeghi Published: 2026-04-12 Categories: cs.AI, cs.HC arXiv: https://arxiv.org/abs/2604.10673v1 AI alignment is often framed as the task of ensuring that an AI system follows a set of stated principles or human preferences, but general principles rarely determine their own application in concrete cases. When principles conflict, when they are too broad to settle a situation, or when the relevant facts are unclear, an additional act of judgment is required. This paper analyzes that step through the lens of hermeneutics and argues that alignment therefore includes an interpretive component: it involves context-sensitive judgments about how principles should be read, applied, and prioritized in practice. We connect this claim to recent empirical findings showing that a substantial portion of preference-labeling data falls into cases of principle conflict or indifference, where the principle set does not uniquely determine a decision. We then draw an operational consequence: because such judgments are expressed in behavior, many alignment-relevant choices appear only in the distribution of responses a model generates at deployment time. To formalize this point, we distinguish deployment-induced and corpus-induced evaluation and show that off-policy audits can fail to capture alignment-relevant failures when the two response distributions differ. We argue that principle-specified alignment includes a context-dependent interpretive component. --- ### One-Step Score-Based Density Ratio Estimation Authors: Wei Chen, Qibin Zhao, John Paisley, Junmei Yang, Delu Zeng Published: 2026-04-12 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2604.10672v1 Density ratio estimation (DRE) is a useful tool for quantifying discrepancies between probability distributions, but existing approaches often involve a trade-off between estimation quality and computational efficiency. Classical direct DRE methods are usually efficient at inference time, yet their performance can seriously deteriorate when the discrepancy between distributions is large. In contrast, score-based DRE methods often yield more accurate estimates in such settings, but they typically require considerable repeated function evaluations and numerical integration. We propose One-step Score-based Density Ratio Estimation (OS-DRE), a partly analytic and solver-free framework designed to combine these complementary advantages. OS-DRE decomposes the time score into spatial and temporal components, representing the latter with an analytic radial basis function (RBF) frame. This formulation converts the otherwise intractable temporal integral into a closed-form weighted sum, thereby removing the need for numerical solvers and enabling DRE with only one function evaluation. We further analyze approximation conditions for the analytic frame, and establish approximation error bounds for both finitely and infinitely smooth temporal kernels, grounding the framework in existing approximation theory. Experiments across density estimation, continual Kullback-Leibler and mutual information estimation, and near out-of-distribution detection demonstrate that OS-DRE offers a favorable balance between estimation quality and inference efficiency. --- ### Learning and Enforcing Context-Sensitive Control for LLMs Authors: Mohammad Albinhassan, Pranava Madhyastha, Mark Law, Alessandra Russo Published: 2026-04-12 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2604.10667v1 Controlling the output of Large Language Models (LLMs) through context-sensitive constraints has emerged as a promising approach to overcome the limitations of Context-Free Grammars (CFGs) in guaranteeing generation validity. However, such constraints typically require manual specification -- a significant barrier demanding specialized expertise. We introduce a framework that automatically learns context-sensitive constraints from LLM interactions through a two-phase process: syntactic exploration to gather diverse outputs for constraint learning, followed by constraint exploitation to enforce these learned rules during generation. Experiments demonstrate that our method enables even small LLMs (1B parameters) to learn and generate with perfect constraint adherence, outperforming larger counterparts and state-of-the-art reasoning models. This work represents the first integration of context-sensitive grammar learning with LLM generation, eliminating manual specification while maintaining generation validity. --- ### Omnimodal Dataset Distillation via High-order Proxy Alignment Authors: Yuxuan Gao, Xiaohao Liu, Xiaobo Xia, Tongliang Liu Published: 2026-04-12 Categories: cs.CV, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2604.10666v1 Dataset distillation compresses large-scale datasets into compact synthetic sets while preserving training performance, but existing methods are largely restricted to single-modal or bimodal settings. Extending dataset distillation to scenarios involving more than two modalities, i.e., Omnimodal Dataset Distillation, remains underexplored and challenging due to increased heterogeneity and complex cross-modal interactions. In this work, we identify the key determinant that bounds the endpoint discrepancy in the omnimodal setting, which is exacerbated with an increasing number of modalities. To this end, we propose HoPA, a unified method that captures high-order cross-modal alignments via a compact proxy, which is compatible with trajectory matching as well. By abstracting omnimodal alignment with a shared similarity structure, our method avoids the combinatorial complexity of pairwise modality modeling and enables scalable joint distillation across heterogeneous modalities. Theoretical analysis from the spectral perspective reveals the rationality of our proposed method against bimodal dataset distillation techniques. Extensive experiments on various benchmarks demonstrate that the proposed method achieves superior compression-performance trade-offs compared to existing competitors. The source code will be publicly released. --- ### HeceTokenizer: A Syllable-Based Tokenization Approach for Turkish Retrieval Authors: Senol Gulgonul Published: 2026-04-12 Categories: cs.CL, cs.IR arXiv: https://arxiv.org/abs/2604.10665v1 HeceTokenizer is a syllable-based tokenizer for Turkish that exploits the deterministic six-pattern phonological structure of the language to construct a closed, out-of-vocabulary (OOV)-free vocabulary of approximately 8,000 unique syllable types. A BERT-tiny encoder (1.5M parameters) is trained from scratch on a subset of Turkish Wikipedia using a masked language modeling objective and evaluated on the TQuAD retrieval benchmark using Recall@5. Combined with a fine-grained chunk-based retrieval strategy, HeceTokenizer achieves 50.3% Recall@5, surpassing the 46.92% reported by a morphology-driven baseline that uses a 200 times larger model. These results suggest that the phonological regularity of Turkish syllables provides a strong and resource-light inductive bias for retrieval tasks. --- ### Preference-Agile Multi-Objective Optimization for Real-time Vehicle Dispatching Authors: Jiahuan Jin, Wenhao Zhao, Rong Qu, Jianfeng Ren, Xinan Chen, Qingfu Zhang, Ruibin Bai Published: 2026-04-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.10664v1 Multi-objective optimization (MOO) has been widely studied in literature because of its versatility in human-centered decision making in real-life applications. Recently, demand for dynamic MOO is fast-emerging due to tough market dynamics that require real-time re-adjustments of priorities for different objectives. However, most existing studies focus either on deterministic MOO problems which are not practical, or non-sequential dynamic MOO decision problems that cannot deal with some real-life complexities. To address these challenges, a preference-agile multi-objective optimization (PAMOO) is proposed in this paper to permit users to dynamically adjust and interactively assign the preferences on the fly. To achieve this, a novel uniform model within a deep reinforcement learning (DRL) framework is proposed that can take as inputs users' dynamic preference vectors explicitly. Additionally, a calibration function is fitted to ensure high quality alignment between the preference vector inputs and the output DRL decision policy. Extensive experiments on challenging real-life vehicle dispatching problems at a container terminal showed that PAMOO obtains superior performance and generalization ability when compared with two most popular MOO methods. Our method presents the first dynamic MOO method for challenging \rev{dynamic sequential MOO decision problems --- ### Energy-Efficient Federated Edge Learning For Small-Scale Datasets in Large IoT Networks Authors: Haihui Xie, Wenkun Wen, Shuwu Chen, Zhaogang Shu, Minghua Xia Published: 2026-04-12 Categories: cs.LG, cs.IT arXiv: https://arxiv.org/abs/2604.10662v1 Large-scale Internet of Things (IoT) networks enable intelligent services such as smart cities and autonomous driving, but often face resource constraints. Collecting heterogeneous sensory data, especially in small-scale datasets, is challenging, and independent edge nodes can lead to inefficient resource utilization and reduced learning performance. To address these issues, this paper proposes a collaborative optimization framework for energy-efficient federated edge learning with small-scale datasets. We first derive an expected learning loss to quantify the relationship between the number of training samples and learning objectives. A stochastic online learning algorithm is then designed to adapt to data variations, and a resource optimization problem with a convergence bound is formulated. Finally, an online distributed algorithm efficiently solves large-scale optimization problems with high scalability. Extensive simulations and autonomous navigation case studies with collision avoidance demonstrate that the proposed approach significantly improves learning performance and resource efficiency compared to state-of-the-art benchmarks. --- ### DynamicsLLM: a Dynamic Analysis-based Tool for Generating Intelligent Execution Traces Using LLMs to Detect Android Behavioural Code Smells Authors: Houcine Abdelkader Cherief, Florent Avellaneda, Naouel Moha Published: 2026-04-12 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2604.10661v1 Mobile apps have become essential of our daily lives, making code quality a critical concern for developers. Behavioural code smells are characteristics in the source code that induce inappropriate code behaviour during execution, which negatively impact software quality in terms of performance, energy consumption, and memory. Dynamics, the latest state-of-the-art tool-based method, is highly effective at detecting Android behavioural code smells. While it outperforms static analysis tools, it suffers from a high false negative rate, with multiple code smell instances remaining undetected. Large Language Models (LLMs) have achieved notable advances across numerous research domains and offer significant potential for generating intelligent execution traces, particularly for detecting behavioural code smells in Android mobile applications. By intelligent execution trace, we mean a sequence of events generated by specific actions in a way that triggers the identification of a given behaviour. We propose the following three main contributions in this paper: (1) DynamicsLLM, an enhanced implementation of the Dynamics method that leverages LLMs to intelligently generate execution traces. (2) A novel hybrid approach designed to improve the coverage of code smell-related events in applications with a small number of activities. (3) A comprehensive validation of DynamicsLLM on 333 mobile applications from F-DROID, including a comparison with the Dynamics tool. Our results show that, under a limited number of actions, DynamicsLLM configured with 100% LLM covers three times more code smell-related events than Dynamics. The hybrid approach improves LLM coverage by 25.9% for apps containing few activities. Moreover, 12.7% of the code smell-related events that cannot be triggered by Dynamics are successfully triggered by our tool. --- ### Efficient Process Reward Modeling via Contrastive Mutual Information Authors: Nakyung Lee, Sangwoo Hong, Jungwoo Lee Published: 2026-04-12 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2604.10660v1 Recent research has devoted considerable effort to verifying the intermediate reasoning steps of chain-of-thought (CoT) trajectories using process reward models (PRMs) and other verifier models. However, training a PRM typically requires human annotators to assign reward scores to each reasoning step, which is both costly and time-consuming. Existing automated approaches, such as Monte Carlo (MC) estimation, also demand substantial computational resources due to repeated LLM rollouts. To overcome these limitations, we propose contrastive pointwise mutual information (CPMI), a novel automatic reward labeling method that leverages the model's internal probability to infer step-level supervision while significantly reducing the computational burden of annotating dataset. CPMI quantifies how much a reasoning step increases the mutual information between the step and the correct target answer relative to hard-negative alternatives. This contrastive signal serves as a proxy for the step's contribution to the final solution and yields a reliable reward. The experimental results show that CPMI-based labeling reduces dataset construction time by 84% and token generation by 98% compared to MC estimation, while achieving higher accuracy on process-level evaluations and mathematical reasoning benchmarks. --- ### Governed Reasoning for Institutional AI Authors: Mamadou Seck Published: 2026-04-12 Categories: cs.AI, cs.CY, cs.MA arXiv: https://arxiv.org/abs/2604.10658v1 Institutional decisions -- regulatory compliance, clinical triage, prior authorization appeal -- require a different AI architecture than general-purpose agents provide. Agent frameworks infer authority conversationally, reconstruct accountability from logs, and produce silent errors: incorrect determinations that execute without any human review signal. We propose Cognitive Core: a governed decision substrate built from nine typed cognitive primitives (retrieve, classify, investigate, verify, challenge, reflect, deliberate, govern, generate), a four-tier governance model where human review is a condition of execution rather than a post-hoc check, a tamper-evident SHA-256 hash-chain audit ledger endogenous to computation, and a demand-driven delegation architecture supporting both declared and autonomously reasoned epistemic sequences. We benchmark three systems on an 11-case balanced prior authorization appeal evaluation set. Cognitive Core achieves 91% accuracy against 55% (ReAct) and 45% (Plan-and-Solve). The governance result is more significant: CC produced zero silent errors while both baselines produced 5-6. We introduce governability -- how reliably a system knows when it should not act autonomously -- as a primary evaluation axis for institutional AI alongside accuracy. The baselines are implemented as prompts, representing the realistic deployment alternative to a governed framework. A configuration-driven domain model means deploying a new institutional decision domain requires YAML configuration, not engineering capacity. --- ### LoViF 2026 The First Challenge on Weather Removal in Videos Authors: Chenghao Qian, Xin Li, Yeying Jin, Shangguan Sun, Yilian Zhong, Yuxiang Chen, Shibo Yin, Yushun Fang, Xilei Zhu, Yahui Wang, Chen Lu, Ying Fu et al. Published: 2026-04-12 Categories: cs.CV, cs.AI, cs.MM arXiv: https://arxiv.org/abs/2604.10655v2 This paper presents a review of the LoViF 2026 Challenge on Weather Removal in Videos. The challenge encourages the development of methods for restoring clean videos from inputs degraded by adverse weather conditions such as rain and snow, with an emphasis on achieving visually plausible and temporally consistent results while preserving scene structure and motion dynamics. To support this task, we introduce a new short-form WRV dataset tailored for video weather removal. It consists of 18 videos 1,216 synthesized frames paired with 1,216 real-world ground-truth frames at a resolution of 832 x 480, and is split into training, validation, and test sets with a ratio of 1:1:1. The goal of this challenge is to advance robust and realistic video restoration under real-world weather conditions, with evaluation protocols that jointly consider fidelity and perceptual quality. The challenge attracted 37 participants and received 5 valid final submissions with corresponding fact sheets, contributing to progress in weather removal for videos. The project is publicly available at https://www.codabench.org/competitions/13462/. --- ### Enhancing Cross-Problem Vehicle Routing via Federated Learning Authors: Xiangchi Meng, Jianan Zhou, Jie Gao, Yifan Lu, Yaoxin Wu, Gonglin Yuan, Yaqing Hou Published: 2026-04-12 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2604.10652v1 Vehicle routing problems (VRPs) constitute a core optimization challenge in modern logistics and supply chain management. The recent neural combinatorial optimization (NCO) has demonstrated superior efficiency over some traditional algorithms. While serving as a primary NCO approach for solving general VRPs, current cross-problem learning paradigms are still subject to performance degradation and generalizability decay, when transferring from simple VRP variants to those involving different and complex constraints. To strengthen the paradigms, this paper offers an innovative "Multi-problem Pre-train, then Single-problem Fine-tune" framework with Federated Learning (MPSF-FL). This framework exploits the common knowledge of a federated global model to foster efficient cross-problem knowledge sharing and transfer among local models for single-problem fine-tuning. In this way, local models effectively retain common VRP knowledge from up-to-date global model, while being efficiently adapted to downstream VRPs with heterogeneous complex constraints. Experimental results demonstrate that our framework not only enhances the performance in diverse VRPs, but also improves the generalizability in unseen problems. --- ### A Deep Generative Approach to Stratified Learning Authors: Randy Martinez, Rong Tang, Lizhen Lin Published: 2026-04-12 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2604.10650v1 While the manifold hypothesis is widely adopted in modern machine learning, complex data is often better modeled as stratified spaces -- unions of manifolds (strata) of varying dimensions. Stratified learning is challenging due to varying dimensionality, intersection singularities, and lack of efficient models in learning the underlying distributions. We provide a deep generative approach to stratified learning by developing two generative frameworks for learning distributions on stratified spaces. The first is a sieve maximum likelihood approach realized via a dimension-aware mixture of variational autoencoders. The second is a diffusion-based framework that explores the score field structure of a mixture. We establish the convergence rates for learning both the ambient and intrinsic distributions, which are shown to be dependent on the intrinsic dimensions and smoothness of the underlying strata. Utilizing the geometry of the score field, we also establish consistency for estimating the intrinsic dimension of each stratum and propose an algorithm that consistently estimates both the number of strata and their dimensions. Theoretical results for both frameworks provide fundamental insights into the interplay of the underlying geometry, the ambient noise level, and deep generative models. Extensive simulations and real dataset applications, such as molecular dynamics, demonstrate the effectiveness of our methods. --- ### SpectralLoRA: Is Low-Frequency Structure Sufficient for LoRA Adaptation? A Spectral Analysis of Weight Updates Authors: Rajveer Singh Published: 2026-04-12 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2604.10649v2 We present a systematic empirical study of the spectral structure of LoRA weight updates. Through 2D Discrete Cosine Transform (DCT) analysis of trained adaptation matrices across BERT-base and RoBERTa-base on four GLUE benchmarks (SST-2, MNLI, CoLA, QQP), we establish that LoRA updates are universally dominated by low-frequency components: on average, just 33% of DCT coefficients capture 90% of total spectral energy. Retaining only 10% of frequency coefficients reduces adapter storage by 10x while sacrificing only 1.95 percentage points on SST-2. Notably, frequency masking at k=50% improves over full LoRA on 3 of 8 model-task pairs, suggesting high-frequency components act as adaptation noise. We further discover that RoBERTa-base is systematically more spectrally compressible than BERT-base across all tasks, and that task complexity governs spectral sensitivity: NLI tasks require more frequency budget than sentiment classification. A subsequent SVD-DCT correlation analysis (Pearson r=0.906, p<1e-9) connects the empirical 33% constant to the spectral dynamics of SGD (Olsen et al., 2025), suggesting a theoretical grounding for this finding. These findings motivate a new design principle for PEFT: spectral sparsity in adaptation. --- ### Vibe-driven model-based engineering Authors: Jordi Cabot Published: 2026-04-12 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2604.10645v2 There is a pressing need for better development methods and tools to keep up with the growing demand and increasing complexity of new software systems. New types of user interfaces, the need for intelligent components, sustainability concerns, etc. bring new challenges that we need to handle. In the last years, model-driven engineering (MDE), including its latest incarnation, i.e. low/no-code development, has been key to improving the quality and productivity of software development, but models themselves are becoming increasingly complex to specify and manage. At the same time, we are witnessing the growing popularity of vibe coding approaches that rely on Large Language Models (LLMs) to transform natural language descriptions into running code at the expense of potential code vulnerabilities, scalability issues and maintainability concerns. While many may think vibe coding will replace model-based engineering, in this paper we argue that, in fact, the two approaches can complement each other and provide altogether different development paths for different types of software systems, development scenarios, and user profiles. In this sense, we introduce the concept of \textit{vibe-driven model-based engineering} as a novel approach to integrate the best of both worlds (AI and MDE) to accelerate the development of reliable complex systems. We outline the key concepts of this new approach and highlight the opportunities and open challenges it presents for the future of software development. ---