# Research Papers — 2026-01 · bundle page 132 of 189 > Full-text markdown of papers #6551–6600 of 9403 published in 2026-01, newest first. AI-readable mirrors of the full 9403-paper 2026-01 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-01/page/132 [← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/131.md) [Next bundle →](https://aipod.dev/papers/archive/2026-01/page/133.md) ### ShowUI-Aloha: Human-Taught GUI Agent Authors: Yichun Zhang, Xiangwu Guo, Yauhong Goh, Jessica Hu, Zhiheng Chen, Xin Wang, Difei Gao, Mike Zheng Shou Published: 2026-01-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.07181v1 Graphical User Interfaces (GUIs) are central to human-computer interaction, yet automating complex GUI tasks remains a major challenge for autonomous agents, largely due to a lack of scalable, high-quality training data. While recordings of human demonstrations offer a rich data source, they are typically long, unstructured, and lack annotations, making them difficult for agents to learn from.To address this, we introduce ShowUI-Aloha, a comprehensive pipeline that transforms unstructured, in-the-wild human screen recordings from desktop environments into structured, actionable tasks. Our framework includes four key components: A recorder that captures screen video along with precise user interactions like mouse clicks, keystrokes, and scrolls. A learner that semantically interprets these raw interactions and the surrounding visual context, translating them into descriptive natural language captions. A planner that reads the parsed demonstrations, maintains task states, and dynamically formulates the next high-level action plan based on contextual reasoning. An executor that faithfully carries out these action plans at the OS level, performing precise clicks, drags, text inputs, and window operations with safety checks and real-time feedback. Together, these components provide a scalable solution for collecting and parsing real-world human data, demonstrating a viable path toward building general-purpose GUI agents that can learn effectively from simply observing humans. --- ### Structured Reasoning for Large Language Models Authors: Jinyi Han, Zixiang Di, Zishang Jiang, Ying Liao, Jiaqing Liang, Yongqi Wang, Yanghua Xiao Published: 2026-01-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.07180v1 Large language models (LLMs) achieve strong performance by generating long chains of thought, but longer traces always introduce redundant or ineffective reasoning steps. One typical behavior is that they often perform unnecessary verification and revisions even if they have reached the correct answers. This limitation stems from the unstructured nature of reasoning trajectories and the lack of targeted supervision for critical reasoning abilities. To address this, we propose Structured Reasoning (SCR), a framework that decouples reasoning trajectories into explicit, evaluable, and trainable components. We mainly implement SCR using a Generate-Verify-Revise paradigm. Specifically, we construct structured training data and apply Dynamic Termination Supervision to guide the model in deciding when to terminate reasoning. To avoid interference between learning signals for different reasoning abilities, we adopt a progressive two-stage reinforcement learning strategy: the first stage targets initial generation and self-verification, and the second stage focuses on revision. Extensive experiments on three backbone models show that SCR substantially improves reasoning efficiency and self-verification. Besides, compared with existing reasoning paradigms, it reduces output token length by up to 50%. --- ### DIVER: Dynamic Iterative Visual Evidence Reasoning for Multimodal Fake News Detection Authors: Weilin Zhou, Zonghao Ying, Chunlei Meng, Jiahui Liu, Hengyang Zhou, Quanchen Zou, Deyue Zhang, Dongdong Yang, Xiangzheng Zhang Published: 2026-01-12 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.07178v1 Multimodal fake news detection is crucial for mitigating adversarial misinformation. Existing methods, relying on static fusion or LLMs, face computational redundancy and hallucination risks due to weak visual foundations. To address this, we propose DIVER (Dynamic Iterative Visual Evidence Reasoning), a framework grounded in a progressive, evidence-driven reasoning paradigm. DIVER first establishes a strong text-based baseline through language analysis, leveraging intra-modal consistency to filter unreliable or hallucinated claims. Only when textual evidence is insufficient does the framework introduce visual information, where inter-modal alignment verification adaptively determines whether deeper visual inspection is necessary. For samples exhibiting significant cross-modal semantic discrepancies, DIVER selectively invokes fine-grained visual tools (e.g., OCR and dense captioning) to extract task-relevant evidence, which is iteratively aggregated via uncertainty-aware fusion to refine multimodal reasoning. Experiments on Weibo, Weibo21, and GossipCop demonstrate that DIVER outperforms state-of-the-art baselines by an average of 2.72\%, while optimizing inference efficiency with a reduced latency of 4.12 s. --- ### Safe-FedLLM: Delving into the Safety of Federated Large Language Models Authors: Mingxiang Tao, Yu Tian, Wenxuan Tu, Yue Yang, Xue Yang, Xiangyan Tang Published: 2026-01-12 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2601.07177v5 Federated learning (FL) addresses privacy and data-silo issues in the training of large language models (LLMs). Most prior work focuses on improving the efficiency of federated learning for LLMs (FedLLM). However, security in open federated environments, particularly defenses against malicious clients, remains underexplored. To investigate the security of FedLLM, we conduct a preliminary study to analyze potential attack surfaces and defensive characteristics from the perspective of LoRA updates. We find two key properties of FedLLM: 1) LLMs are vulnerable to attacks from malicious clients in FL, and 2) LoRA updates exhibit distinct behavioral patterns that can be effectively distinguished by lightweight classifiers. Based on these properties, we propose Safe-FedLLM, a probe-based defense framework for FedLLM, which constructs defenses across three levels: Step-Level, Client-Level, and Shadow-Level. The core concept of Safe-FedLLM is to perform probe-based discrimination on each client's local LoRA updates, treating them as high-dimensional behavioral features and using a lightweight classifier to determine whether they are malicious. Extensive experiments demonstrate that Safe-FedLLM effectively improves FedLLM's robustness against malicious clients while maintaining competitive performance on benign data. Notably, our method effectively suppresses the impact of malicious data without significantly affecting training speed, and remains effective even under high malicious client ratios. --- ### Offline Meta-Reinforcement Learning with Flow-Based Task Inference and Adaptive Correction of Feature Overgeneralization Authors: Min Wang, Xin Li, Mingzhong Wang, Hasnaa Bennis Published: 2026-01-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.07164v1 Offline meta-reinforcement learning (OMRL) combines the strengths of learning from diverse datasets in offline RL with the adaptability to new tasks of meta-RL, promising safe and efficient knowledge acquisition by RL agents. However, OMRL still suffers extrapolation errors due to out-of-distribution (OOD) actions, compromised by broad task distributions and Markov Decision Process (MDP) ambiguity in meta-RL setups. Existing research indicates that the generalization of the $Q$ network affects the extrapolation error in offline RL. This paper investigates this relationship by decomposing the $Q$ value into feature and weight components, observing that while decomposition enhances adaptability and convergence in the case of high-quality data, it often leads to policy degeneration or collapse in complex tasks. We observe that decomposed $Q$ values introduce a large estimation bias when the feature encounters OOD samples, a phenomenon we term ''feature overgeneralization''. To address this issue, we propose FLORA, which identifies OOD samples by modeling feature distributions and estimating their uncertainties. FLORA integrates a return feedback mechanism to adaptively adjust feature components. Furthermore, to learn precise task representations, FLORA explicitly models the complex task distribution using a chain of invertible transformations. We theoretically and empirically demonstrate that FLORA achieves rapid adaptation and meta-policy improvement compared to baselines across various environments. --- ### Test-time Adaptive Hierarchical Co-enhanced Denoising Network for Reliable Multimodal Classification Authors: Shu Shen, C. L. Philip Chen, Tong Zhang Published: 2026-01-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.07163v2 Reliable learning of multimodal data (e.g., multi-omics) is a widely concerning issue, especially in safety-critical applications such as medical diagnosis. However, low-quality data induced by multimodal noise poses a major challenge in this domain, causing existing methods to suffer from two key limitations. First, they struggle to handle heterogeneous data noise, hindering robust multimodal representation learning. Second, they exhibit limited adaptability and generalization when encountering previously unseen noise. To address these issues, we propose Test-time Adaptive Hierarchical Co-enhanced Denoising Network (TAHCD). On one hand, TAHCD introduces the Adaptive Stable Subspace Alignment and Sample-Adaptive Confidence Alignment to reliably remove heterogeneous noise. They account for noise at both global and instance levels and enable jointly removal of modality-specific and cross-modality noise, achieving robust learning. On the other hand, TAHCD introduces Test-Time Cooperative Enhancement, which adaptively updates the model in response to input noise in a label-free manner, thus improving generalization. This is achieved by collaboratively enhancing the joint removal process of modality-specific and cross-modality noise across global and instance levels according to sample noise. Experiments on multiple benchmarks demonstrate that the proposed method achieves superior classification performance, robustness, and generalization compared with state-of-the-art reliable multimodal learning approaches. --- ### AscendKernelGen: A Systematic Study of LLM-Based Kernel Generation for Neural Processing Units Authors: Xinzi Cao, Jianyang Zhai, Pengfei Li, Zhiheng Hu, Cen Yan, Bingxu Mu, Guanghuan Fang, Bin She, Jiayu Li, Yihan Su, Dongyang Tao, Xiansong Huang et al. Published: 2026-01-12 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.07160v2 To meet the ever-increasing demand for computational efficiency, Neural Processing Units (NPUs) have become critical in modern AI infrastructure. However, unlocking their full potential requires developing high-performance compute kernels using vendor-specific Domain-Specific Languages (DSLs), a task that demands deep hardware expertise and is labor-intensive. While Large Language Models (LLMs) have shown promise in general code generation, they struggle with the strict constraints and scarcity of training data in the NPU domain. Our preliminary study reveals that state-of-the-art general-purpose LLMs fail to generate functional complex kernels for Ascend NPUs, yielding a near-zero success rate. To address these challenges, we propose AscendKernelGen, a generation-evaluation integrated framework for NPU kernel development. We introduce Ascend-CoT, a high-quality dataset incorporating chain-of-thought reasoning derived from real-world kernel implementations, and KernelGen-LM, a domain-adaptive model trained via supervised fine-tuning and reinforcement learning with execution feedback. Furthermore, we design NPUKernelBench, a comprehensive benchmark for assessing compilation, correctness, and performance across varying complexity levels. Experimental results demonstrate that our approach significantly bridges the gap between general LLMs and hardware-specific coding. Specifically, the compilation success rate on complex Level-2 kernels improves from 0% to 95.5% (Pass@10), while functional correctness achieves 64.3% compared to the baseline's complete failure. These results highlight the critical role of domain-specific reasoning and rigorous evaluation in automating accelerator-aware code generation. AscendKernGen is available at https://huggingface.co/AscendKernelGen and https://github.com/weich97/NPUKernelBench. --- ### Stable On-Policy Distillation through Adaptive Target Reformulation Authors: Ijun Jang, Jewon Yeom, Juan Yeo, Hyunggyu Lim, Taesup Kim Published: 2026-01-12 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.07155v3 Knowledge distillation (KD) is a widely adopted technique for transferring knowledge from large language models to smaller student models; however, conventional supervised KD often suffers from a distribution mismatch between training and inference. While on-policy KD approaches attempt to mitigate this issue by learning directly from student-generated outputs, they frequently encounter training instabilities because the distributional gap between the novice student and the expert teacher is often too wide to bridge directly. These challenges manifest as pathological gradients in forward KL objectives or diversity collapse in reverse KL regimes. To address these limitations, we propose Veto, an objective-level reformulation that constructs a geometric bridge in the logit space. Unlike prior methods that mix data samples, Veto creates an intermediate target distribution that promotes alignment between the teacher and the student. By introducing a tunable parameter beta, Veto serves as an Adaptive Gradient Veto that stabilizes optimization by suppressing harmful gradients on low-confidence tokens, while simultaneously acting as a Decisiveness Knob to balance reward-driven performance with output diversity. Extensive experiments across various reasoning and generation tasks demonstrate that Veto consistently outperforms supervised fine-tuning and existing on-policy baselines. --- ### Motion Focus Recognition in Fast-Moving Egocentric Video Authors: Si-En Hong, James Tribble, Alexander Lake, Hao Wang, Chaoyi Zhou, Ashish Bastola, Siyu Huang, Eisa Chaudhary, Brian Canada, Ismahan Arslan-Ari, Abolfazl Razi Published: 2026-01-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.07154v3 From Vision-Language-Action (VLA) systems to robotics, existing egocentric datasets primarily focus on action recognition tasks, while largely overlooking the inherent role of motion analysis in sports and other fast-movement scenarios. To bridge this gap, we propose a real-time motion focus recognition method that estimates the subject's locomotion intention from any egocentric video. We leverage the foundation model for camera pose estimation and introduce system-level optimizations to enable efficient and scalable inference. Evaluated on a collected egocentric action dataset, our method achieves real-time performance with manageable memory consumption through a sliding batch inference strategy. This work makes motion-centric analysis practical for edge deployment and offers a complementary perspective to existing egocentric studies on sports and fast-movement activities. --- ### Can Large Language Models Understand, Reason About, and Generate Code-Switched Text? Authors: Genta Indra Winata, David Anugraha, Patrick Amadeus Irawan, Anirban Das, Haneul Yoo, Paresh Dashore, Shreyas Kulkarni, Ruochen Zhang, Haruki Sakajo, Frederikus Hudi, Anaelia Ovalle, Syrielle Montariol et al. Published: 2026-01-12 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.07153v1 Code-switching is a pervasive phenomenon in multilingual communication, yet the robustness of large language models (LLMs) in mixed-language settings remains insufficiently understood. In this work, we present a comprehensive evaluation of LLM capabilities in understanding, reasoning over, and generating code-switched text. We introduce CodeMixQA a novel benchmark with high-quality human annotations, comprising 16 diverse parallel code-switched language-pair variants that span multiple geographic regions and code-switching patterns, and include both original scripts and their transliterated forms. Using this benchmark, we analyze the reasoning behavior of LLMs on code-switched question-answering tasks, shedding light on how models process and reason over mixed-language inputs. We further conduct a systematic evaluation of LLM-generated synthetic code-switched text, focusing on both naturalness and semantic fidelity, and uncover key limitations in current generation capabilities. Our findings reveal persistent challenges in both reasoning and generation under code-switching conditions and provide actionable insights for building more robust multilingual LLMs. We release the dataset and code as open source. --- ### Rewarding Creativity: A Human-Aligned Generative Reward Model for Reinforcement Learning in Storytelling Authors: Zhaoyan Li, Hang Lei, Yujia Wang, Lanbo Liu, Hao Liu, Liang Yu Published: 2026-01-12 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2601.07149v1 While Large Language Models (LLMs) can generate fluent text, producing high-quality creative stories remains challenging. Reinforcement Learning (RL) offers a promising solution but faces two critical obstacles: designing reliable reward signals for subjective storytelling quality and mitigating training instability. This paper introduces the Reinforcement Learning for Creative Storytelling (RLCS) framework to systematically address both challenges. First, we develop a Generative Reward Model (GenRM) that provides multi-dimensional analysis and explicit reasoning about story preferences, trained through supervised fine-tuning on demonstrations with reasoning chains distilled from strong teacher models, followed by GRPO-based refinement on expanded preference data. Second, we introduce an entropy-based reward shaping strategy that dynamically prioritizes learning on confident errors and uncertain correct predictions, preventing overfitting on already-mastered patterns. Experiments demonstrate that GenRM achieves 68\% alignment with human creativity judgments, and RLCS significantly outperforms strong baselines including Gemini-2.5-Pro in overall story quality. This work provides a practical pipeline for applying RL to creative domains, effectively navigating the dual challenges of reward modeling and training stability. --- ### Measuring Iterative Temporal Reasoning with Time Puzzles Authors: Zhengxiang Wang, Zeyu Dong Published: 2026-01-12 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.07148v3 Tool use, such as web search, has become a standard capability even in freely available large language models (LLMs). However, existing benchmarks evaluate temporal reasoning mainly in static, non-tool-using settings, which poorly reflect how LLMs perform temporal reasoning in practice. We introduce Time Puzzles, a constraint-based date inference task for evaluating iterative temporal reasoning with tools. Each puzzle combines factual temporal anchors with (cross-cultural) calendar relations and may admit one or multiple valid dates. The puzzles are algorithmically generated, enabling controlled and continual evaluation. Across 13 LLMs, even the best model (GPT-5) achieves only 55.3% accuracy without tools, despite using easily searchable facts. While web search improves performance, models perform substantially better when constraints are rewritten with explicit dates, removing the need for factual lookup. These results reveal a gap in reliable tool use for iterative temporal reasoning. --- ### Generating readily synthesizable small molecule fluorophore scaffolds with reinforcement learning Authors: Ruhi Sayana, Kate Callon, Jennifer Xu, Jonathan Deutsch, Steven Chu, James Zou, John Janetzko, Rabindra V. Shivnaraine, Kyle Swanson Published: 2026-01-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.07145v1 Developing new fluorophores for advanced imaging techniques requires exploring new chemical space. While generative AI approaches have shown promise in designing novel dye scaffolds, prior efforts often produced synthetically intractable candidates due to a lack of reaction constraints. Here, we developed SyntheFluor-RL, a generative AI model that employs known reaction libraries and molecular building blocks to create readily synthesizable fluorescent molecule scaffolds via reinforcement learning. To guide the generation of fluorophores, SyntheFluor-RL employs a scoring function built on multiple graph neural networks (GNNs) that predict key photophysical properties, including photoluminescence quantum yield, absorption, and emission wavelengths. These outputs are dynamically weighted and combined with a computed pi-conjugation score to prioritize candidates with desirable optical characteristics and synthetic feasibility. SyntheFluor-RL generated 11,590 candidate molecules, which were filtered to 19 structures predicted to possess dye-like properties. Of the 19 molecules, 14 were synthesized and 13 were experimentally confirmed. The top three were characterized, with the lead compound featuring a benzothiadiazole chromophore and exhibiting strong fluorescence (PLQY = 0.62), a large Stokes shift (97 nm), and a long excited-state lifetime (11.5 ns). These results demonstrate the effectiveness of SyntheFluor-RL in the identification of synthetically accessible fluorophores for further development. --- ### Optimal Transport under Group Fairness Constraints Authors: Linus Bleistein, Mathieu Dagréou, Francisco Andrade, Thomas Boudou, Aurélien Bellet Published: 2026-01-12 Categories: stat.ML, cs.LG, math.ST arXiv: https://arxiv.org/abs/2601.07144v3 Ensuring fairness in matching algorithms is a key challenge in allocating scarce resources and positions. Focusing on Optimal Transport (OT), we introduce a novel notion of group fairness requiring that the probability of matching two individuals from any two given groups in the OT plan satisfies a predefined target. We first propose a modified Sinkhorn algorithm to compute perfectly fair transport plans efficiently. Since exact fairness can significantly degrade matching quality in practice, we then develop two relaxation strategies. The first one involves solving a penalized OT problem, for which we derive novel finite-sample complexity guarantees. Our second strategy leverages bilevel optimization to learn a ground cost that induces a fair OT solution, and we establish a bound on the deviation of fairness when matching unseen data. Finally, we present empirical results illustrating the performance of our approaches and the trade-off between fairness and transport cost. --- ### A Large-Scale Study on the Development and Issues of Multi-Agent AI Systems Authors: Daniel Liu, Krishna Upadhyay, Vinaik Chhetri, A. B. Siddique, Umar Farooq Published: 2026-01-12 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2601.07136v1 The rapid emergence of multi-agent AI systems (MAS), including LangChain, CrewAI, and AutoGen, has shaped how large language model (LLM) applications are developed and orchestrated. However, little is known about how these systems evolve and are maintained in practice. This paper presents the first large-scale empirical study of open-source MAS, analyzing over 42K unique commits and over 4.7K resolved issues across eight leading systems. Our analysis identifies three distinct development profiles: sustained, steady, and burst-driven. These profiles reflect substantial variation in ecosystem maturity. Perfective commits constitute 40.8% of all changes, suggesting that feature enhancement is prioritized over corrective maintenance (27.4%) and adaptive updates (24.3%). Data about issues shows that the most frequent concerns involve bugs (22%), infrastructure (14%), and agent coordination challenges (10%). Issue reporting also increased sharply across all frameworks starting in 2023. Median resolution times range from under one day to about two weeks, with distributions skewed toward fast responses but a minority of issues requiring extended attention. These results highlight both the momentum and the fragility of the current ecosystem, emphasizing the need for improved testing infrastructure, documentation quality, and maintenance practices to ensure long-term reliability and sustainability. --- ### Proof of Reasoning for Privacy Enhanced Federated Blockchain Learning at the Edge Authors: James Calo, Benny Lo Published: 2026-01-12 Categories: cs.CR, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.07134v1 Consensus mechanisms are the core of any blockchain system. However, the majority of these mechanisms do not target federated learning directly nor do they aid in the aggregation step. This paper introduces Proof of Reasoning (PoR), a novel consensus mechanism specifically designed for federated learning using blockchain, aimed at preserving data privacy, defending against malicious attacks, and enhancing the validation of participating networks. Unlike generic blockchain consensus mechanisms commonly found in the literature, PoR integrates three distinct processes tailored for federated learning. Firstly, a masked autoencoder (MAE) is trained to generate an encoder that functions as a feature map and obfuscates input data, rendering it resistant to human reconstruction and model inversion attacks. Secondly, a downstream classifier is trained at the edge, receiving input from the trained encoder. The downstream network's weights, a single encoded datapoint, the network's output and the ground truth are then added to a block for federated aggregation. Lastly, this data facilitates the aggregation of all participating networks, enabling more complex and verifiable aggregation methods than previously possible. This three-stage process results in more robust networks with significantly reduced computational complexity, maintaining high accuracy by training only the downstream classifier at the edge. PoR scales to large IoT networks with low latency and storage growth, and adapts to evolving data, regulations, and network conditions. --- ### The Potential Impact of Neuromorphic Computing on Radio Telescope Observatories Authors: Nicholas J. Pritchard, Richard Dodson, Andreas Wicenec Published: 2026-01-12 Categories: astro-ph.IM, cs.NE arXiv: https://arxiv.org/abs/2601.07130v1 Radio astronomy relies on bespoke, experimental and innovative computing solutions. This will continue as next-generation telescopes such as the Square Kilometre Array (SKA) and next-generation Very Large Array (ngVLA) take shape. Under increasingly demanding power consumption, and increasingly challenging radio environments, science goals may become intractable with conventional von Neumann computing due to related power requirements. Neuromorphic computing offers a compelling alternative, and combined with a desire for data-driven methods, Spiking Neural Networks (SNNs) are a promising real-time power-efficient alternative. Radio Frequency Interference (RFI) detection is an attractive use-case for SNNs where recent exploration holds promise. This work presents a comprehensive analysis of the potential impact of deploying varying neuromorphic approaches across key stages in radio astronomy processing pipelines for several existing and near-term instruments. Our analysis paves a realistic path from near-term FPGA deployment of SNNs in existing instruments, allowing the addition of advanced data-driven RFI detection for no capital cost, to neuromorphic ASICs for future instruments, finding that commercially available solutions could reduce the power budget for key processing elements by up to three orders of magnitude, transforming the operational budget of the observatory. High-data-rate spectrographic processing could be a well-suited target for the neuromorphic computing industry, as we cast radio telescopes as the world's largest in-sensor compute challenge. --- ### ReinPool: Reinforcement Learning Pooling Multi-Vector Embeddings for Retrieval System Authors: Sungguk Cha, DongWook Kim, Mintae Kim, Youngsub Han, Byoung-Ki Jeon, Sangyeob Lee Published: 2026-01-12 Categories: cs.IR, cs.CL, cs.CV arXiv: https://arxiv.org/abs/2601.07125v1 Multi-vector embedding models have emerged as a powerful paradigm for document retrieval, preserving fine-grained visual and textual details through token-level representations. However, this expressiveness comes at a staggering cost: storing embeddings for every token inflates index sizes by over $1000\times$ compared to single-vector approaches, severely limiting scalability. We introduce \textbf{ReinPool}, a reinforcement learning framework that learns to dynamically filter and pool multi-vector embeddings into compact, retrieval-optimized representations. By training with an inverse retrieval objective and NDCG-based rewards, ReinPool identifies and retains only the most discriminative vectors without requiring manual importance annotations. On the Vidore V2 benchmark across three vision-language embedding models, ReinPool compresses multi-vector representations by $746$--$1249\times$ into single vectors while recovering 76--81\% of full multi-vector retrieval performance. Compared to static mean pooling baselines, ReinPool achieves 22--33\% absolute NDCG@3 improvement, demonstrating that learned selection significantly outperforms heuristic aggregation. --- ### Towards Automated Diagnosis of Inherited Arrhythmias: Combined Arrhythmia Classification Using Lead-Aware Spatial Attention Networks Authors: Sophie Sigfstead, River Jiang, Brianna Davies, Zachary W. M. Laksman, Julia Cadrin-Tourigny, Rafik Tadros, Habib Khan, Joseph Atallah, Christian Steinberg, Shubhayan Sanatani, Mario Talajic, Rahul Krishnan et al. Published: 2026-01-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.07124v1 Arrhythmogenic right ventricular cardiomyopathy (ARVC) and long QT syndrome (LQTS) are inherited arrhythmia syndromes associated with sudden cardiac death. Deep learning shows promise for ECG interpretation, but multi-class inherited arrhythmia classification with clinically grounded interpretability remains underdeveloped. Our objective was to develop and validate a lead-aware deep learning framework for multi-class (ARVC vs LQTS vs control) and binary inherited arrhythmia classification, and to determine optimal strategies for integrating ECG foundation models within arrhythmia screening tools. We assembled a 13-center Canadian cohort (645 patients; 1,344 ECGs). We evaluated four ECG foundation models using three transfer learning approaches: linear probing, fine-tuning, and combined strategies. We developed lead-aware spatial attention networks (LASAN) and assessed integration strategies combining LASAN with foundation models. Performance was compared against the established foundation model baselines. Lead-group masking quantified disease-specific lead dependence. Fine-tuning outperformed linear probing and combined strategies across all foundation models (mean macro-AUROC 0.904 vs 0.825). The best lead-aware integrations achieved near-ceiling performance (HuBERT-ECG hybrid: macro-AUROC 0.990; ARVC vs control AUROC 0.999; LQTS vs control AUROC 0.994). Lead masking demonstrated physiologic plausibility: V1-V3 were most critical for ARVC detection (4.54% AUROC reduction), while lateral leads were preferentially important for LQTS (2.60% drop). Lead-aware architectures achieved state-of-the-art performance for inherited arrhythmia classification, outperforming all existing published models on both binary and multi-class tasks while demonstrating clinically aligned lead dependence. These findings support potential utility for automated ECG screening pending validation. --- ### ENTRA: Entropy-Based Redundancy Avoidance in Large Language Model Reasoning Authors: Ruichu Cai, Haopeng Du, Qingwen Lin, Yutong Chen, Zijian Li, Boyan Xu Published: 2026-01-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.07123v1 Large Reasoning Models (LRMs) often suffer from overthinking, generating unnecessarily long reasoning chains even for simple tasks. This leads to substantial computational overhead with limited performance gain, primarily due to redundant verification and repetitive generation. While prior work typically constrains output length or optimizes correctness, such coarse supervision fails to guide models toward concise yet accurate inference. In this paper, we propose ENTRA, an entropy-based training framework that suppresses redundant reasoning while preserving performance. ENTRA first estimates the token-level importance using a lightweight Bidirectional Importance Estimation (BIE) method, which accounts for both prediction confidence and forward influence. It then computes a redundancy reward based on the entropy of low-importance tokens, normalized by its theoretical upper bound, and optimizes this reward via reinforcement learning. Experiments on mathematical reasoning benchmarks demonstrate that ENTRA reduces output length by 37% to 53% with no loss-and in some cases, gains-in accuracy. Our approach offers a principled and efficient solution to reduce overthinking in LRMs, and provides a generalizable path toward redundancy-aware reasoning optimization. --- ### Enhancing Cloud Network Resilience via a Robust LLM-Empowered Multi-Agent Reinforcement Learning Framework Authors: Yixiao Peng, Hao Hu, Feiyang Li, Xinye Cao, Yingchang Jiang, Jipeng Tang, Guoshun Nan, Yuling Liu Published: 2026-01-12 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.07122v2 While virtualization and resource pooling empower cloud networks with structural flexibility and elastic scalability, they inevitably expand the attack surface and challenge cyber resilience. Reinforcement Learning (RL)-based defense strategies have been developed to optimize resource deployment and isolation policies under adversarial conditions, aiming to enhance system resilience by maintaining and restoring network availability. However, existing approaches lack robustness as they require retraining to adapt to dynamic changes in network structure, node scale, attack strategies, and attack intensity. Furthermore, the lack of Human-in-the-Loop (HITL) support limits interpretability and flexibility. To address these limitations, we propose CyberOps-Bots, a hierarchical multi-agent reinforcement learning framework empowered by Large Language Models (LLMs). Inspired by MITRE ATT&CK's Tactics-Techniques model, CyberOps-Bots features a two-layer architecture: (1) An upper-level LLM agent with four modules--ReAct planning, IPDRR-based perception, long-short term memory, and action/tool integration--performs global awareness, human intent recognition, and tactical planning; (2) Lower-level RL agents, developed via heterogeneous separated pre-training, execute atomic defense actions within localized network regions. This synergy preserves LLM adaptability and interpretability while ensuring reliable RL execution. Experiments on real cloud datasets show that, compared to state-of-the-art algorithms, CyberOps-Bots maintains network availability 68.5% higher and achieves a 34.7% jumpstart performance gain when shifting the scenarios without retraining. To our knowledge, this is the first study to establish a robust LLM-RL framework with HITL support for cloud defense. --- ### Enhancing Cloud Network Resilience via a Robust LLM-Empowered Multi-Agent Reinforcement Learning Framework Authors: Yixiao Peng, Hao Hu, Feiyang Li, Xinye Cao, Yingchang Jiang, Jipeng Tang, Guoshun Nan, Yuling Liu Published: 2026-01-12 Categories: cs.CR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.07122v1 While virtualization and resource pooling empower cloud networks with structural flexibility and elastic scalability, they inevitably expand the attack surface and challenge cyber resilience. Reinforcement Learning (RL)-based defense strategies have been developed to optimize resource deployment and isolation policies under adversarial conditions, aiming to enhance system resilience by maintaining and restoring network availability. However, existing approaches lack robustness as they require retraining to adapt to dynamic changes in network structure, node scale, attack strategies, and attack intensity. Furthermore, the lack of Human-in-the-Loop (HITL) support limits interpretability and flexibility. To address these limitations, we propose CyberOps-Bots, a hierarchical multi-agent reinforcement learning framework empowered by Large Language Models (LLMs). Inspired by MITRE ATT&CK's Tactics-Techniques model, CyberOps-Bots features a two-layer architecture: (1) An upper-level LLM agent with four modules--ReAct planning, IPDRR-based perception, long-short term memory, and action/tool integration--performs global awareness, human intent recognition, and tactical planning; (2) Lower-level RL agents, developed via heterogeneous separated pre-training, execute atomic defense actions within localized network regions. This synergy preserves LLM adaptability and interpretability while ensuring reliable RL execution. Experiments on real cloud datasets show that, compared to state-of-the-art algorithms, CyberOps-Bots maintains network availability 68.5% higher and achieves a 34.7% jumpstart performance gain when shifting the scenarios without retraining. To our knowledge, this is the first study to establish a robust LLM-RL framework with HITL support for cloud defense. We will release our framework to the community, facilitating the advancement of robust and autonomous defense in cloud networks. --- ### ReMIND: Orchestrating Modular Large Language Models for Controllable Serendipity A REM-Inspired System Design for Emergent Creative Ideation Authors: Makoto Sato Published: 2026-01-12 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.07121v3 Large language models (LLMs) are increasingly used not only for problem solving but also for creative ideation; however, generating ideas that are both novel and coherent remains challenging. While high-temperature sampling can promote originality, it often compromises consistency and usefulness. Here, we propose ReMIND, a four-stage framework comprising wake, which establishes a stable semantic baseline through low-temperature generation; dream, which performs high-temperature exploratory generation; judge, which evaluates candidate outputs for consistency and extracts salient novel ideas; and rewake, which consolidates selected ideas into coherent final outputs. By assigning these functions to independent LLM modules, ReMIND explicitly separates exploration from stabilization. We systematically evaluated diverse model configurations using multiple creative ideation tasks. External evaluations showed that novelty enhancement could emerge through two separable stages: first from the wake to the dream phase through high-temperature exploration, and subsequently from the dream to the rewake phase through judge-mediated selection and consolidation. Notably, different LLM families exhibited distinct functional tendencies. These findings suggest that serendipitous ideation in LLMs is not solely a property of individual models but emerges from interactions among models assigned to specialized cognitive roles. ReMIND provides a general framework for investigating computational creativity and demonstrates how modular LLM orchestration can bridge exploratory generation and coherent idea formation. --- ### ReMIND: Orchestrating Modular Large Language Models for Controllable Serendipity A REM-Inspired System Design for Emergent Creative Ideation Authors: Makoto Sato Published: 2026-01-12 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.07121v1 Large language models (LLMs) are used not only for problem solving but also for creative ideation; however, eliciting serendipitous insights that are both novel and internally coherent remains difficult. While stochastic sampling promotes novelty, it often degrades consistency. Here, we propose ReMIND, a REM-inspired modular framework for ideation. ReMIND consists of four stages: wake, which generates a stable low-temperature semantic baseline; dream, which performs high-temperature exploratory generation; judge, which applies coarse evaluation to filter incoherent outputs and extract candidate ideas; and re-wake, which re-articulates selected ideas into coherent final outputs. By instantiating each stage as an independent LLM, ReMIND enables functional separation between exploration and consolidation. Parameter sweeps show that ReMIND reliably induces semantic exploration while preserving downstream stability. Embedding-based analyses confirm substantial semantic displacement during the dream phase, whereas external evaluations reveal that high-quality ideas emerge sporadically rather than as extrema along any single metric. These results suggest that serendipitous ideation in LLMs is a rare-event process best approached through system level design that shapes the conditions under which valuable ideas can emerge and be stabilized. ReMIND provides a general framework for studying the computational basis of serendipity and illustrates how modular LLM orchestration can bridge exploration and stabilization. --- ### SC-MII: Infrastructure LiDAR-based 3D Object Detection on Edge Devices for Split Computing with Multiple Intermediate Outputs Integration Authors: Taisuke Noguchi, Takayuki Nishio, Takuya Azumi Published: 2026-01-12 Categories: cs.DC, cs.CV arXiv: https://arxiv.org/abs/2601.07119v1 3D object detection using LiDAR-based point cloud data and deep neural networks is essential in autonomous driving technology. However, deploying state-of-the-art models on edge devices present challenges due to high computational demands and energy consumption. Additionally, single LiDAR setups suffer from blind spots. This paper proposes SC-MII, multiple infrastructure LiDAR-based 3D object detection on edge devices for Split Computing with Multiple Intermediate outputs Integration. In SC-MII, edge devices process local point clouds through the initial DNN layers and send intermediate outputs to an edge server. The server integrates these features and completes inference, reducing both latency and device load while improving privacy. Experimental results on a real-world dataset show a 2.19x speed-up and a 71.6% reduction in edge device processing time, with at most a 1.09% drop in accuracy. --- ### Reward-Preserving Attacks For Robust Reinforcement Learning Authors: Lucas Schott, Elies Gherbi, Hatem Hajri, Sylvain Lamprier Published: 2026-01-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.07118v2 Adversarial training in reinforcement learning (RL) is challenging because perturbations cascade through trajectories and compound over time, making fixed-strength attacks either overly destructive or too conservative. We propose reward-preserving attacks, which adapt adversarial strength so that an $α$ fraction of the nominal-to-worst-case return gap remains achievable at each state. In deep RL, perturbation magnitudes $η$ are selected dynamically, using a learned critic $Q((s,a),η)$ that estimates the expected return of $α$-reward-preserving rollouts. For intermediate values of $α$, this adaptive training yields policies that are robust across a wide range of perturbation magnitudes while preserving nominal performance, outperforming fixed-radius and uniformly sampled-radius adversarial training. --- ### Reward-Preserving Attacks For Robust Reinforcement Learning Authors: Lucas Schott, Elies Gherbi, Hatem Hajri, Sylvain Lamprier Published: 2026-01-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.07118v1 Adversarial robustness in RL is difficult because perturbations affect entire trajectories: strong attacks can break learning, while weak attacks yield little robustness, and the appropriate strength varies by state. We propose $α$-reward-preserving attacks, which adapt the strength of the adversary so that an $α$ fraction of the nominal-to-worst-case return gap remains achievable at each state. In deep RL, we use a gradient-based attack direction and learn a state-dependent magnitude $η\le η_{\mathcal B}$ selected via a critic $Q^π_α((s,a),η)$ trained off-policy over diverse radii. This adaptive tuning calibrates attack strength and, with intermediate $α$, improves robustness across radii while preserving nominal performance, outperforming fixed- and random-radius baselines. --- ### Few-shot Class-Incremental Learning via Generative Co-Memory Regularization Authors: Kexin Bao, Yong Li, Dan Zeng, Shiming Ge Published: 2026-01-12 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.07117v1 Few-shot class-incremental learning (FSCIL) aims to incrementally learn models from a small amount of novel data, which requires strong representation and adaptation ability of models learned under few-example supervision to avoid catastrophic forgetting on old classes and overfitting to novel classes. This work proposes a generative co-memory regularization approach to facilitate FSCIL. In the approach, the base learning leverages generative domain adaptation finetuning to finetune a pretrained generative encoder on a few examples of base classes by jointly incorporating a masked autoencoder (MAE) decoder for feature reconstruction and a fully-connected classifier for feature classification, which enables the model to efficiently capture general and adaptable representations. Using the finetuned encoder and learned classifier, we construct two class-wise memories: representation memory for storing the mean features for each class, and weight memory for storing the classifier weights. After that, the memory-regularized incremental learning is performed to train the classifier dynamically on the examples of few-shot classes in each incremental session by simultaneously optimizing feature classification and co-memory regularization. The memories are updated in a class-incremental manner and they collaboratively regularize the incremental learning. In this way, the learned models improve recognition accuracy, while mitigating catastrophic forgetting over old classes and overfitting to novel classes. Extensive experiments on popular benchmarks clearly demonstrate that our approach outperforms the state-of-the-arts. --- ### The Need for a Socially-Grounded Persona Framework for User Simulation Authors: Pranav Narayanan Venkit, Yu Li, Yada Pruksachatkun, Chien-Sheng Wu Published: 2026-01-12 Categories: cs.CL, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2601.07110v2 Synthetic personas are widely used to condition large language models (LLMs) for social simulation, yet most personas are still constructed from coarse sociodemographic attributes or summaries. We revisit persona creation by introducing SCOPE, a socially grounded framework for persona construction and evaluation, built from a 141-item, two-hour sociopsychological protocol collected from 124 U.S.-based participants. Across seven models, we find that demographic-only personas are a structural bottleneck: demographics explain only ~1.5% of variance in human response similarity. Adding sociopsychological facets improves behavioral prediction and reduces over-accentuation, and non-demographic personas based on values and identity achieve strong alignment with substantially lower bias. These trends generalize to SimBench (441 aligned questions), where SCOPE personas outperform default prompting and NVIDIA Nemotron personas, and SCOPE augmentation improves Nemotron-based personas. Our results indicate that persona quality depends on sociopsychological structure rather than demographic templates or summaries. --- ### The Need for a Socially-Grounded Persona Framework for User Simulation Authors: Pranav Narayanan Venkit, Yu Li, Yada Pruksachatkun, Chien-Sheng Wu Published: 2026-01-12 Categories: cs.CL, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2601.07110v1 Synthetic personas are widely used to condition large language models (LLMs) for social simulation, yet most personas are still constructed from coarse sociodemographic attributes or summaries. We revisit persona creation by introducing SCOPE, a socially grounded framework for persona construction and evaluation, built from a 141-item, two-hour sociopsychological protocol collected from 124 U.S.-based participants. Across seven models, we find that demographic-only personas are a structural bottleneck: demographics explain only ~1.5% of variance in human response similarity. Adding sociopsychological facets improves behavioral prediction and reduces over-accentuation, and non-demographic personas based on values and identity achieve strong alignment with substantially lower bias. These trends generalize to SimBench (441 aligned questions), where SCOPE personas outperform default prompting and NVIDIA Nemotron personas, and SCOPE augmentation improves Nemotron-based personas. Our results indicate that persona quality depends on sociopsychological structure rather than demographic templates or summaries. --- ### MEDVISTAGYM: A Scalable Training Environment for Thinking with Medical Images via Tool-Integrated Reinforcement Learning Authors: Meng Lu, Yuxing Lu, Yuchen Zhuang, Megan Mullins, Yang Xie, Guanghua Xiao, Charles Fleming, Wenqi Shi, Xuan Wang Published: 2026-01-12 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.07107v1 Vision language models (VLMs) achieve strong performance on general image understanding but struggle to think with medical images, especially when performing multi-step reasoning through iterative visual interaction. Medical VLMs often rely on static visual embeddings and single-pass inference, preventing models from re-examining, verifying, or refining visual evidence during reasoning. While tool-integrated reasoning offers a promising path forward, open-source VLMs lack the training infrastructure to learn effective tool selection, invocation, and coordination in multi-modal medical reasoning. We introduce MedVistaGym, a scalable and interactive training environment that incentivizes tool-integrated visual reasoning for medical image analysis. MedVistaGym equips VLMs to determine when and which tools to invoke, localize task-relevant image regions, and integrate single or multiple sub-image evidence into interleaved multimodal reasoning within a unified, executable interface for agentic training. Using MedVistaGym, we train MedVistaGym-R1 to interleave tool use with agentic reasoning through trajectory sampling and end-to-end reinforcement learning. Across six medical VQA benchmarks, MedVistaGym-R1-8B exceeds comparably sized tool-augmented baselines by 19.10% to 24.21%, demonstrating that structured agentic training--not tool access alone--unlocks effective tool-integrated reasoning for medical image analysis. --- ### Assessing LLM Response Quality in the Context of Technology-Facilitated Abuse Authors: Vijay Prakash, Majed Almansoori, Donghan Hu, Rahul Chatterjee, Danny Yuxing Huang Published: 2026-01-11 Categories: cs.HC, cs.AI, cs.CL, cs.CR, cs.CY arXiv: https://arxiv.org/abs/2602.17672v1 Technology-facilitated abuse (TFA) is a pervasive form of intimate partner violence (IPV) that leverages digital tools to control, surveil, or harm survivors. While tech clinics are one of the reliable sources of support for TFA survivors, they face limitations due to staffing constraints and logistical barriers. As a result, many survivors turn to online resources for assistance. With the growing accessibility and popularity of large language models (LLMs), and increasing interest from IPV organizations, survivors may begin to consult LLM-based chatbots before seeking help from tech clinics. In this work, we present the first expert-led manual evaluation of four LLMs - two widely used general-purpose non-reasoning models and two domain-specific models designed for IPV contexts - focused on their effectiveness in responding to TFA-related questions. Using real-world questions collected from literature and online forums, we assess the quality of zero-shot single-turn LLM responses generated with a survivor safety-centered prompt on criteria tailored to the TFA domain. Additionally, we conducted a user study to evaluate the perceived actionability of these responses from the perspective of individuals who have experienced TFA. Our findings, grounded in both expert assessment and user feedback, provide insights into the current capabilities and limitations of LLMs in the TFA context and may inform the design, development, and fine-tuning of future models for this domain. We conclude with concrete recommendations to improve LLM performance for survivor support. --- ### AI Hallucination from Students' Perspective: A Thematic Analysis Authors: Abdulhadi Shoufan, Ahmad-Azmi-Abdelhamid Esmaeil Published: 2026-01-11 Categories: cs.HC, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2602.17671v1 As students increasingly rely on large language models, hallucinations pose a growing threat to learning. To mitigate this, AI literacy must expand beyond prompt engineering to address how students should detect and respond to LLM hallucinations. To support this, we need to understand how students experience hallucinations, how they detect them, and why they believe they occur. To investigate these questions, we asked university students three open-ended questions about their experiences with AI hallucinations, their detection strategies, and their mental models of why hallucinations occur. Sixty-three students responded to the survey. Thematic analysis of their responses revealed that reported hallucination issues primarily relate to incorrect or fabricated citations, false information, overconfident but misleading responses, poor adherence to prompts, persistence in incorrect answers, and sycophancy. To detect hallucinations, students rely either on intuitive judgment or on active verification strategies, such as cross-checking with external sources or re-prompting the model. Students' explanations for why hallucinations occur reflected several mental models, including notable misconceptions. Many described AI as a research engine that fabricates information when it cannot locate an answer in its "database." Others attributed hallucinations to issues with training data, inadequate prompting, or the model's inability to understand or verify information. These findings illuminate vulnerabilities in AI-supported learning and highlight the need for explicit instruction in verification protocols, accurate mental models of generative AI, and awareness of behaviors such as sycophancy and confident delivery that obscure inaccuracy. The study contributes empirical evidence for integrating hallucination awareness and mitigation into AI literacy curricula. --- ### SDUs DAISY: A Benchmark for Danish Culture Authors: Jacob Nielsen, Stine L. Beltoft, Peter Schneider-Kamp, Lukas Galke Poech Published: 2026-01-11 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.19930v2 We introduce Daisy, a factual knowledge benchmark for Danish cultural heritage, based on curated topics from the Danish Culture Canon 2006. For each artifact in the culture canon, we query the corresponding Wikipedia page and have a language model generate a diverse set of questions. Within each artifact, we sample both central and peripheral questions, testing not only mainstream information but also the deeper, defining elements of Danish cultural heritage as identified by the Canon committee. Each question-answer pair is manually approved or corrected, yielding a final dataset of 741 closed-ended question-answer pairs ranging from archaeological findings dated to 1300 BCE and 18th-century poems and musical pieces through to contemporary pop music, Danish design, and architecture. Baseline results on our benchmark show that contemporary language models (GPT-OSS-120B & 20B, Llama-3.3-70B, Gemma3-27B, and Mistral-3.1-24B) perform surprisingly poorly on this task, despite the tested factual knowledge being available on Wikipedia. The best-performing model, Llama-3.3-70B, attains only 0.17 BLEU and an F1 score of 0.27, demonstrating the benchmark's difficulty. The questions, model benchmarks, and the evaluation tool are publicly available at https://github.com/schneiderkamplab/SDU-Daisy and under schneiderkamplab/SDU-Daisy on Hugging Face Datasets. --- ### Stingy Context: 18:1 Hierarchical Code Compression for LLM Auto-Coding Authors: David Linus Ostby Published: 2026-01-11 Categories: cs.CL, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2601.19929v2 We introduce Stingy Context, a hierarchical tree-based compression scheme achieving 18:1 reduction in LLM context for auto-coding tasks. Using our TREEFRAG exploit decomposition, we reduce a real source code base of 239k tokens to 11k tokens while preserving task fidelity. Empirical results across 12 Frontier models show 94 to 97% success on 40 real-world issues at low cost, outperforming flat methods and mitigating lost-in-the-middle effects. --- ### Towards a Mechanistic Understanding of Large Reasoning Models: A Survey of Training, Inference, and Failures Authors: Yi Hu, Jiaqi Gu, Ruxin Wang, Zijun Yao, Hao Peng, Xiaobao Wu, Jianhui Chen, Muhan Zhang, Liangming Pan Published: 2026-01-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.19928v1 Reinforcement learning (RL) has catalyzed the emergence of Large Reasoning Models (LRMs) that have pushed reasoning capabilities to new heights. While their performance has garnered significant excitement, exploring the internal mechanisms driving these behaviors has become an equally critical research frontier. This paper provides a comprehensive survey of the mechanistic understanding of LRMs, organizing recent findings into three core dimensions: 1) training dynamics, 2) reasoning mechanisms, and 3) unintended behaviors. By synthesizing these insights, we aim to bridge the gap between black-box performance and mechanistic transparency. Finally, we discuss under-explored challenges to outline a roadmap for future mechanistic studies, including the need for applied interpretability, improved methodologies, and a unified theoretical framework. --- ### Which Quantization Should I Use? A Unified Evaluation of llama.cpp Quantization on Llama-3.1-8B-Instruct Authors: Uygar Kurt Published: 2026-01-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.14277v1 Quantization is a practical technique for making large language models easier to deploy by reducing the precision used to store and operate on model weights. This can lower memory use and improve runtime feasibility on constrained hardware, which is especially relevant for users running models locally. Quantization in llama.cpp enables large language models to run on commodity hardware, but available formats are often evaluated inconsistently, making it hard to choose among schemes. We present a unified empirical study of the llama.cpp quantization on a single modern model, Llama-3.1-8B-Instruct (FP16, GGUF), covering 3-8 bit K-quant and legacy formats. We evaluate downstream task performance across standard reasoning, knowledge, instruction-following, and truthfulness benchmarks, and also measure perplexity and CPU throughput (prefill/decoding) alongside model size, compression, and quantization time. Ultimately, this work is a practical guide for choosing a llama.cpp quantization scheme, helping readers make informed, context-aware decisions for their intended use and resource budget. --- ### Dynamical Systems Analysis Reveals Functional Regimes in Large Language Models Authors: Hassan Ugail, Newton Howard Published: 2026-01-11 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.11622v1 Large language models perform text generation through high-dimensional internal dynamics, yet the temporal organisation of these dynamics remains poorly understood. Most interpretability approaches emphasise static representations or causal interventions, leaving temporal structure largely unexplored. Drawing on neuroscience, where temporal integration and metastability are core markers of neural organisation, we adapt these concepts to transformer models and discuss a composite dynamical metric, computed from activation time-series during autoregressive generation. We evaluate this metric in GPT-2-medium across five conditions: structured reasoning, forced repetition, high-temperature noisy sampling, attention-head pruning, and weight-noise injection. Structured reasoning consistently exhibits elevated metric relative to repetitive, noisy, and perturbed regimes, with statistically significant differences confirmed by one-way ANOVA and large effect sizes in key comparisons. These results are robust to layer selection, channel subsampling, and random seeds. Our findings demonstrate that neuroscience-inspired dynamical metrics can reliably characterise differences in computational organisation across functional regimes in large language models. We stress that the proposed metric captures formal dynamical properties and does not imply subjective experience. --- ### A Mind Cannot Be Smeared Across Time Authors: Michael Timothy Bennett Published: 2026-01-11 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.11620v2 Whether machines can be conscious depends not only on what they compute, but \emph{when} they compute it. Most deployed artificial systems realise their functions via sequential or time-multiplexed updates, yet a moment of conscious experience feels unified and simultaneous. I prove that this difference matters. I augment Stack Theory with algebraic laws relating within time-window constraint satisfaction to conjunction. I introduce a temporal semantics over windowed trajectories $τ_Δ$ and prove that existential temporal realisation $\Diamond_Δ$ does not preserve conjunction. A system can realise all the ingredients of experience across time without ever instantiating the experienced conjunction itself. I then distinguish two postulates, Chord and Arpeggio. Chord is the position that conscious unity requires \textit{objective co-instantiation} of the grounded conjunction within the window, like a musical chord. Arpeggio only needs the ingredients to \textit{occur} within window, like a melody. I formalise concurrency-capacity to measure what is needed to satisfy co-instantiation. Finally, I review neurophysiological evidence suggesting that consciousness depends on phase synchrony and effective connectivity, and that loss of consciousness is associated with its breakdown. Under Chord, software consciousness on strictly sequential substrates is impossible for contents whose grounding requires two or more simultaneous contributors. The hardware matters. --- ### Explore with Long-term Memory: A Benchmark and Multimodal LLM-based Reinforcement Learning Framework for Embodied Exploration Authors: Sen Wang, Bangwei Liu, Zhenkun Gao, Lizhuang Ma, Xuhong Wang, Yuan Xie, Xin Tan Published: 2026-01-11 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2601.10744v2 An ideal embodied agent should possess lifelong learning capabilities to handle long-horizon and complex tasks, enabling continuous operation in general environments. This not only requires the agent to accurately accomplish given tasks but also to leverage long-term episodic memory to optimize decision-making. However, existing mainstream one-shot embodied tasks primarily focus on task completion results, neglecting the crucial process of exploration and memory utilization. To address this, we propose Long-term Memory Embodied Exploration (LMEE), which aims to unify the agent's exploratory cognition and decision-making behaviors to promote lifelong learning. We further construct a corresponding dataset and benchmark, LMEE-Bench, incorporating multi-goal navigation and memory-based question answering to comprehensively evaluate both the process and outcome of embodied exploration. To enhance the agent's memory recall and proactive exploration capabilities, we propose MemoryExplorer, a novel method that fine-tunes a multimodal large language model through reinforcement learning to encourage active memory querying. By incorporating a multi-task reward function that includes action prediction, frontier selection, and question answering, our model achieves proactive exploration. Extensive experiments against state-of-the-art embodied exploration models demonstrate that our approach achieves significant advantages in long-horizon embodied tasks. Our dataset and code will be released at https://wangsen99.github.io/papers/lmee/ --- ### UBiGTLoc: A Unified BiLSTM-Graph Transformer Localization Framework for IoT Sensor Networks Authors: Ayesh Abu Lehyeh, Anastassia Gharib, Tian Xia, Dryver Huston, Safwan Wshah Published: 2026-01-11 Categories: eess.SP, cs.LG, cs.NI arXiv: https://arxiv.org/abs/2601.10743v1 Sensor nodes localization in wireless Internet of Things (IoT) sensor networks is crucial for the effective operation of diverse applications, such as smart cities and smart agriculture. Existing sensor nodes localization approaches heavily rely on anchor nodes within wireless sensor networks (WSNs). Anchor nodes are sensor nodes equipped with global positioning system (GPS) receivers and thus, have known locations. These anchor nodes operate as references to localize other sensor nodes. However, the presence of anchor nodes may not always be feasible in real-world IoT scenarios. Additionally, localization accuracy can be compromised by fluctuations in Received Signal Strength Indicator (RSSI), particularly under non-line-of-sight (NLOS) conditions. To address these challenges, we propose UBiGTLoc, a Unified Bidirectional Long Short-Term Memory (BiLSTM)-Graph Transformer Localization framework. The proposed UBiGTLoc framework effectively localizes sensor nodes in both anchor-free and anchor-presence WSNs. The framework leverages BiLSTM networks to capture temporal variations in RSSI data and employs Graph Transformer layers to model spatial relationships between sensor nodes. Extensive simulations demonstrate that UBiGTLoc consistently outperforms existing methods and provides robust localization across both dense and sparse WSNs while relying solely on cost-effective RSSI data. --- ### Enhancing Formal Software Specification with Artificial Intelligence Authors: Antonio Abu Nassar, Eitan Farchi Published: 2026-01-11 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2601.09745v1 Formal software specification is known to enable early error detection and explicit invariants, yet it has seen limited industrial adoption due to its high notation overhead and the expertise required to use traditional formal languages. This paper presents a case study showing that recent advances in artificial intelligence make it possible to retain many of the benefits of formal specification while substantially reducing these costs. The necessity of a clear distinction between what is controlled by the system analyst and can highly benefits from the rigor of formal specification and what need not be controlled is demonstrated. We use natural language augmented with lightweight mathematical notation and written in \LaTeX\ as an intermediate specification language, which is reviewed and refined by AI prior to code generation. Applied to a nontrivial simulation of organizational knowledge growth, this approach enables early validation, explicit invariants, and correctness by design, while significantly reducing development effort and producing a correct implementation on the first attempt. --- ### First African Digital Humanism Summer School 2025 Authors: Carine P. Mukamakuza, Monika Lanzenberger, George Metakides, Tim Brown, Hannes Werthner Published: 2026-01-11 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2601.08870v1 Artificial intelligence (AI) has become a transformative force across global societies, reshaping the ways we communicate, collaborate, and make decisions. Yet, as AI systems increasingly mediate interactions between humans, questions about the ability to take into account and understand culture, language, and context have taken center stage. This book explores these questions through a series of articles that try to assess AI's capacity to navigate cross-cultural, multilingual, and high-stakes policy environments, emphasizing human-centered approaches that balance technological innovation with social equity. It brings together six case studies from the First African Digital Humanism Summer School that took place in Kigali, Rwanda in July 2025. --- ### AI Deployment Authorisation: A Global Standard for Machine-Readable Governance of High-Risk Artificial Intelligence Authors: Daniel Djan Saparning Published: 2026-01-11 Categories: cs.CY, cs.AI arXiv: https://arxiv.org/abs/2601.08869v1 Modern artificial intelligence governance lacks a formal, enforceable mechanism for determining whether a given AI system is legally permitted to operate in a specific domain and jurisdiction. Existing tools such as model cards, audits, and benchmark evaluations provide descriptive information about model behavior and training data but do not produce binding deployment decisions with legal or financial force. This paper introduces the AI Deployment Authorisation Score (ADAS), a machine-readable regulatory framework that evaluates AI systems across five legally and economically grounded dimensions: risk, alignment, externality, control, and auditability. ADAS produces a cryptographically verifiable deployment certificate that regulators, insurers, and infrastructure operators can consume as a license to operate, using public-key verification and transparency mechanisms adapted from secure software supply chain and certificate transparency systems. The paper presents the formal specification, decision logic, evidence model, and policy architecture of ADAS and demonstrates how it operationalizes the European Union Artificial Intelligence Act, United States critical infrastructure governance, and insurance underwriting requirements by compiling statutory and regulatory obligations into machine-executable deployment gates. We argue that deployment-level authorization, rather than model-level evaluation, constitutes the missing institutional layer required for safe, lawful, and economically scalable artificial intelligence. --- ### Residual Cross-Modal Fusion Networks for Audio-Visual Navigation Authors: Yi Wang, Yinfeng Yu, Bin Ren Published: 2026-01-11 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2601.08868v1 Audio-visual embodied navigation aims to enable an agent to autonomously localize and reach a sound source in unseen 3D environments by leveraging auditory cues. The key challenge of this task lies in effectively modeling the interaction between heterogeneous features during multimodal fusion, so as to avoid single-modality dominance or information degradation, particularly in cross-domain scenarios. To address this, we propose a Cross-Modal Residual Fusion Network, which introduces bidirectional residual interactions between audio and visual streams to achieve complementary modeling and fine-grained alignment, while maintaining the independence of their representations. Unlike conventional methods that rely on simple concatenation or attention gating, CRFN explicitly models cross-modal interactions via residual connections and incorporates stabilization techniques to improve convergence and robustness. Experiments on the Replica and Matterport3D datasets demonstrate that CRFN significantly outperforms state-of-the-art fusion baselines and achieves stronger cross-domain generalization. Notably, our experiments also reveal that agents exhibit differentiated modality dependence across different datasets. The discovery of this phenomenon provides a new perspective for understanding the cross-modal collaboration mechanism of embodied agents. --- ### R$^2$BD: A Reconstruction-Based Method for Generalizable and Efficient Detection of Fake Images Authors: Qingyu Liu, Zhongjie Ba, Jianmin Guo, Qiu Wang, Zhibo Wang, Jie Shi, Kui Ren Published: 2026-01-11 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.08867v1 Recently, reconstruction-based methods have gained attention for AIGC image detection. These methods leverage pre-trained diffusion models to reconstruct inputs and measure residuals for distinguishing real from fake images. Their key advantage lies in reducing reliance on dataset-specific artifacts and improving generalization under distribution shifts. However, they are limited by significant inefficiency due to multi-step inversion and reconstruction, and their reliance on diffusion backbones further limits generalization to other generative paradigms such as GANs. In this paper, we propose a novel fake image detection framework, called R$^2$BD, built upon two key designs: (1) G-LDM, a unified reconstruction model that simulates the generation behaviors of VAEs, GANs, and diffusion models, thereby broadening the detection scope beyond prior diffusion-only approaches; and (2) a residual bias calculation module that distinguishes real and fake images in a single inference step, which is a significant efficiency improvement over existing methods that typically require 20$+$ steps. Extensive experiments on the benchmark from 10 public datasets demonstrate that R$^2$BD is over 22$\times$ faster than existing reconstruction-based methods while achieving superior detection accuracy. In cross-dataset evaluations, it outperforms state-of-the-art methods by an average of 13.87\%, showing strong efficiency and generalization across diverse generative methods. The code and dataset used for evaluation are available at https://github.com/QingyuLiu/RRBD. --- ### Tackling Heterogeneity in Quantum Federated Learning: An Integrated Sporadic-Personalized Approach Authors: Ratun Rahman, Shaba Shaon, Dinh C. Nguyen Published: 2026-01-11 Categories: quant-ph, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.07882v1 Quantum federated learning (QFL) emerges as a powerful technique that combines quantum computing with federated learning to efficiently process complex data across distributed quantum devices while ensuring data privacy in quantum networks. Despite recent research efforts, existing QFL frameworks struggle to achieve optimal model training performance primarily due to inherent heterogeneity in terms of (i) quantum noise where current quantum devices are subject to varying levels of noise due to varying device quality and susceptibility to quantum decoherence, and (ii) heterogeneous data distributions where data across participating quantum devices are naturally non-independent and identically distributed (non-IID). To address these challenges, we propose a novel integrated sporadic-personalized approach called SPQFL that simultaneously handles quantum noise and data heterogeneity in a single QFL framework. It is featured in two key aspects: (i) for quantum noise heterogeneity, we introduce a notion of sporadic learning to tackle quantum noise heterogeneity across quantum devices, and (ii) for quantum data heterogeneity, we implement personalized learning through model regularization to mitigate overfitting during local training on non-IID quantum data distributions, thereby enhancing the convergence of the global model. Moreover, we conduct a rigorous convergence analysis for the proposed SPQFL framework, with both sporadic and personalized learning considerations. Theoretical findings reveal that the upper bound of the SPQFL algorithm is strongly influenced by both the number of quantum devices and the number of quantum noise measurements. Extensive simulation results in real-world datasets also illustrate that the proposed SPQFL approach yields significant improvements in terms of training performance and convergence stability compared to the state-of-the-art methods. --- ### Sola-Visibility-ISPM: Benchmarking Agentic AI for Identity Security Posture Management Visibility Authors: Gal Engelberg, Konstantin Koutsyi, Leon Goldberg, Reuven Elezra, Idan Pinto, Tal Moalem, Shmuel Cohen, Yoni Weintrob Published: 2026-01-11 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2601.07880v1 Identity Security Posture Management (ISPM) is a core challenge for modern enterprises operating across cloud and SaaS environments. Answering basic ISPM visibility questions, such as understanding identity inventory and configuration hygiene, requires interpreting complex identity data, motivating growing interest in agentic AI systems. Despite this interest, there is currently no standardized way to evaluate how well such systems perform ISPM visibility tasks on real enterprise data. We introduce the Sola Visibility ISPM Benchmark, the first benchmark designed to evaluate agentic AI systems on foundational ISPM visibility tasks using a live, production-grade identity environment spanning AWS, Okta, and Google Workspace. The benchmark focuses on identity inventory and hygiene questions and is accompanied by the Sola AI Agent, a tool-using agent that translates natural-language queries into executable data exploration steps and produces verifiable, evidence-backed answers. Across 77 benchmark questions, the agent achieves strong overall performance, with an expert accuracy of 0.84 and a strict success rate of 0.77. Performance is highest on AWS hygiene tasks, where expert accuracy reaches 0.94, while results on Google Workspace and Okta hygiene tasks are more moderate, yet competitive. Overall, this work provides a practical and reproducible benchmark for evaluating agentic AI systems in identity security and establishes a foundation for future ISPM benchmarks covering more advanced identity analysis and governance tasks. --- ### Sliced-Wasserstein Distribution Alignment Loss Improves the Ultra-Low-Bit Quantization of Large Language Models Authors: Deyu Cao, Yixin Yin, Samin Aref Published: 2026-01-11 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2601.07878v1 The benefits of most large language models come with steep and often hidden economic and environmental costs due to their resource usage inefficiency during deployment. Model quantization improves energy and memory efficiency through representing model parameters by lower-precision values. However, compression below 4-bits often distorts activation distributions and degrades performance. We address this challenge by introducing a sliced Wasserstein loss function for distribution-aware calibration in ultra-low-bit post-training quantization. The proposed loss aligns the output distributions of full-precision and quantized models under random linear projections, complementing standard mean-squared error loss without adding any computational overhead during inference. Our proposed loss function can be incorporated with any post-training quantization framework that has a retraining component. We demonstrate the performance gains of our proposed model by incorporating it with two frontier methods known as OmniQuant and TesseraQ. Compared to these two baselines, the proposed loss consistently improves both perplexity and downstream task accuracy across multiple ultra-low-bit settings. Our proposed loss function recovers 4.12-20.37% of the OmniQuant's lost accuracy on the language model LLaMA-2-7B, 0.93-7.65% on OPT-6.7B, and 2.26-6.20% on LLaMA-2-13B. TesseraQ's accuracy degradation is recovered by 3.63-7.63% in relative terms when augmented by our proposed loss function. Taken together, these results demonstrate that distributional alignment provides a simple yet effective performance boost that can push the limits of frontier quantization methods. Our method is available on GitHub to facilitate future progress in ultra-low-bit quantization. --- ### E^2-LLM: Bridging Neural Signals and Interpretable Affective Analysis Authors: Fei Ma, Han Lin, Yifan Xie, Hongwei Ren, Xiaoyu Shen, Wenbo Ding, Qi Tian Published: 2026-01-11 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.07877v1 Emotion recognition from electroencephalography (EEG) signals remains challenging due to high inter-subject variability, limited labeled data, and the lack of interpretable reasoning in existing approaches. While recent multimodal large language models (MLLMs) have advanced emotion analysis, they have not been adapted to handle the unique spatiotemporal characteristics of neural signals. We present E^2-LLM (EEG-to-Emotion Large Language Model), the first MLLM framework for interpretable emotion analysis from EEG. E^2-LLM integrates a pretrained EEG encoder with Qwen-based LLMs through learnable projection layers, employing a multi-stage training pipeline that encompasses emotion-discriminative pretraining, cross-modal alignment, and instruction tuning with chain-of-thought reasoning. We design a comprehensive evaluation protocol covering basic emotion prediction, multi-task reasoning, and zero-shot scenario understanding. Experiments on the dataset across seven emotion categories demonstrate that E^2-LLM achieves excellent performance on emotion classification, with larger variants showing enhanced reliability and superior zero-shot generalization to complex reasoning scenarios. Our work establishes a new paradigm combining physiological signals with LLM reasoning capabilities, showing that model scaling improves both recognition accuracy and interpretable emotional understanding in affective computing. ---