# Research Papers — 2026-02 · bundle page 109 of 210 > Full-text markdown of papers #5401–5450 of 10493 published in 2026-02, newest first. AI-readable mirrors of the full 10493-paper 2026-02 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-02/page/109 [← Previous bundle](https://aipod.dev/papers/archive/2026-02/page/108.md) [Next bundle →](https://aipod.dev/papers/archive/2026-02/page/110.md) ### Multimodal Fact-Level Attribution for Verifiable Reasoning Authors: David Wan, Han Wang, Ziyang Wang, Elias Stengel-Eskin, Hyunji Lee, Mohit Bansal Published: 2026-02-12 Categories: cs.CL, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2602.11509v2 Multimodal large language models (MLLMs) are increasingly used for real-world tasks involving multi-step reasoning and long-form generation, where reliability requires grounding model outputs in heterogeneous input sources and verifying individual factual claims. However, existing multimodal grounding benchmarks and evaluation methods focus on simplified, observation-based scenarios or limited modalities and fail to assess attribution in complex multimodal reasoning. We introduce MuRGAt (Multimodal Reasoning with Grounded Attribution), a benchmark for evaluating fact-level multimodal attribution in settings that require reasoning beyond direct observation. Given inputs spanning video, audio, and other modalities, MuRGAt requires models to generate answers with explicit reasoning and precise citations, where each citation specifies both modality and temporal segments. To enable reliable assessment, we introduce an automatic evaluation framework that strongly correlates with human judgments. Benchmarking with human and automated scores reveals that even strong MLLMs frequently hallucinate citations despite correct reasoning. Moreover, we observe a key trade-off: increasing reasoning depth or enforcing structured grounding often degrades accuracy, highlighting a significant gap between internal reasoning and verifiable attribution. --- ### RooflineBench: A Benchmarking Framework for On-Device LLMs via Roofline Analysis Authors: Zhen Bi, Xueshu Chen, Luoyang Sun, Yuhang Yao, Qing Shen, Jungang Lou, Cheng Deng Published: 2026-02-12 Categories: cs.LG, cs.AI, cs.AR, cs.PF arXiv: https://arxiv.org/abs/2602.11506v4 The transition toward localized intelligence through Small Language Models (SLMs) has intensified the need for rigorous performance characterization on resource-constrained edge hardware. However, objectively measuring the theoretical performance ceilings of diverse architectures across heterogeneous platforms remains a formidable challenge. In this work, we propose a systematic framework based on the Roofline model that unifies architectural primitives and hardware constraints through the lens of operational intensity (OI). By defining an inference-potential region, we introduce the Relative Inference Potential as a novel metric to compare efficiency differences between Large Language Models (LLMs) on the same hardware substrate. Extensive empirical analysis across diverse compute tiers reveals that variations in performance and OI are significantly influenced by sequence length. We further identify a critical regression in OI as model depth increases. Additionally, our findings highlight an efficiency trap induced by hardware heterogeneity and demonstrate how structural refinements, such as Multi-head Latent Attention (MLA), can effectively unlock latent inference potential across various hardware substrates. These insights provide actionable directions for hardware-software co-design to align neural structures with physical constraints in on-device intelligence. The released code is available in the Appendix C. --- ### Calibrating an Imperfect Auxiliary Predictor for Unobserved No-Purchase Choice Authors: Jiangkai Xiong, Kalyan Talluri, Hanzhao Wang Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11505v2 Firms typically cannot observe key consumer actions: whether customers buy from a competitor, choose not to buy, or even fully consider the firm's offer. This missing outside-option information makes market-size and preference estimation difficult even in simple multinomial logit (MNL) models, and it is a central obstacle in practice when only transaction data are recorded. Existing approaches often rely on auxiliary market-share, aggregated, or cross-market data. We study a complementary setting in which a black-box auxiliary predictor provides outside-option probabilities, but is potentially biased or miscalibrated because it was trained in a different channel, period, or population, or produced by an external machine-learning system. We develop calibration methods that turn such imperfect predictions into statistically valid no-purchase estimates using purchase-only data from the focal environment. First, under affine miscalibration in logit space, we show that a simple regression identifies outside-option utility parameters and yields consistent recovery of no-purchase probabilities without collecting new labels for no-purchase events. Second, under a weaker nearly monotone condition, we propose a rank-based calibration method and derive finite-sample error bounds that cleanly separate auxiliary-predictor quality from first-stage utility-learning error over observed in-set choices. Our analysis also translates estimation error into downstream decision quality for assortment optimization, quantifying how calibration accuracy affects revenue performance. The bounds provide explicit dependence on predictor alignment and utility-learning error, clarifying when each source dominates. Numerical experiments demonstrate improvements in no-purchase estimation and downstream assortment decisions, and we discuss robust aggregation extensions for combining multiple auxiliary predictors. --- ### A Generic Framework for Fair Consensus Clustering in Streams Authors: Diptarka Chakraborty, Kushagra Chatterjee, Debarati Das, Tien-Long Nguyen Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11500v1 Consensus clustering seeks to combine multiple clusterings of the same dataset, potentially derived by considering various non-sensitive attributes by different agents in a multi-agent environment, into a single partitioning that best reflects the overall structure of the underlying dataset. Recent work by Chakraborty et al, introduced a fair variant under proportionate fairness and obtained a constant-factor approximation by naively selecting the best closest fair input clustering; however, their offline approach requires storing all input clusterings, which is prohibitively expensive for most large-scale applications. In this paper, we initiate the study of fair consensus clustering in the streaming model, where input clusterings arrive sequentially and memory is limited. We design the first constant-factor algorithm that processes the stream while storing only a logarithmic number of inputs. En route, we introduce a new generic algorithmic framework that integrates closest fair clustering with cluster fitting, yielding improved approximation guarantees not only in the streaming setting but also when revisited offline. Furthermore, the framework is fairness-agnostic: it applies to any fairness definition for which an approximately close fair clustering can be computed efficiently. Finally, we extend our methods to the more general k-median consensus clustering problem. --- ### What if Agents Could Imagine? Reinforcing Open-Vocabulary HOI Comprehension through Generation Authors: Zhenlong Yuan, Yue Wang, Dapeng Zhang, Kejin Cui, Rui Chen, Jing Tang, Lei Sun, Hongwei Yu, Chengxuan Qian, Xiangxiang Chu, Shuo Li, Yuyin Zhou Published: 2026-02-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2602.11499v2 Multimodal Large Language Models have shown promising capabilities in bridging visual and textual reasoning, yet their reasoning capabilities in Open-Vocabulary Human-Object Interaction (OV-HOI) are limited by cross-modal hallucinations and limited viewpoints of images. To address this, we propose ImagineAgent, an agentic framework that integrates cognitive mapping, tool-augmented reinforcement learning (RL), and generative world modeling for robust OV-HOI understanding. Specifically, we first propose an innovative CoT dataset named hicodet-6K for supervised fine-tuning (SFT), which effectively bridges the perception-to-cognition gap by structuring perceived entities into interaction pairs for comprehensive predictions. Subsequently, we develop a multimodal tool library integrating online retrieval, image cropping, and generative modeling, enabling the agent to dynamically augment reasoning with domain-specific tools to resolve visual-semantic ambiguities and hallucinations during inference. Moreover, we incorporate a generative model to reconstruct alternative viewpoints, enabling the agent to 'imagine' under limited viewpoints. Finally, we propose a composite reward mechanism to jointly optimize prediction accuracy and tool efficiency. Evaluations on both SWIG-HOI and HICO-DET datasets demonstrate that our method achieves state-of-the-art performance while requiring merely 36.7% of the training data compared to existing methods, validating our robustness, empirical effectiveness and efficiency. --- ### Partial GFlowNet: Accelerating Convergence in Large State Spaces via Strategic Partitioning Authors: Xuan Yu, Xu Wang, Rui Zhu, Yudong Zhang, Yang Wang Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11498v2 Generative Flow Networks (GFlowNets) have shown promising potential to generate high-scoring candidates with probability proportional to their rewards. As existing GFlowNets freely explore in state space, they encounter significant convergence challenges when scaling to large state spaces. Addressing this issue, this paper proposes to restrict the exploration of actor. A planner is introduced to partition the entire state space into overlapping partial state spaces. Given their limited size, these partial state spaces allow the actor to efficiently identify subregions with higher rewards. A heuristic strategy is introduced to switch partial regions thus preventing the actor from wasting time exploring fully explored or low-reward partial regions. By iteratively exploring these partial state spaces, the actor learns to converge towards the high-reward subregions within the entire state space. Experiments on several widely used datasets demonstrate that \modelname converges faster than existing works on large state spaces. Furthermore, \modelname not only generates candidates with higher rewards but also significantly improves their diversity. --- ### Partial GFlowNet: Accelerating Convergence in Large State Spaces via Strategic Partitioning Authors: Xuan Yu, Xu Wang, Rui Zhu, Yudong Zhang, Yang Wang Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11498v1 Generative Flow Networks (GFlowNets) have shown promising potential to generate high-scoring candidates with probability proportional to their rewards. As existing GFlowNets freely explore in state space, they encounter significant convergence challenges when scaling to large state spaces. Addressing this issue, this paper proposes to restrict the exploration of actor. A planner is introduced to partition the entire state space into overlapping partial state spaces. Given their limited size, these partial state spaces allow the actor to efficiently identify subregions with higher rewards. A heuristic strategy is introduced to switch partial regions thus preventing the actor from wasting time exploring fully explored or low-reward partial regions. By iteratively exploring these partial state spaces, the actor learns to converge towards the high-reward subregions within the entire state space. Experiments on several widely used datasets demonstrate that \modelname converges faster than existing works on large state spaces. Furthermore, \modelname not only generates candidates with higher rewards but also significantly improves their diversity. --- ### Jailbreaking Leaves a Trace: Understanding and Detecting Jailbreak Attacks from Internal Representations of Large Language Models Authors: Sri Durga Sai Sowmya Kadali, Evangelos E. Papalexakis Published: 2026-02-12 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2602.11495v2 Jailbreaking large language models (LLMs) has emerged as a critical security challenge with the widespread deployment of conversational AI systems. Adversarial users exploit these models through carefully crafted prompts to elicit restricted or unsafe outputs, a phenomenon commonly referred to as Jailbreaking. Despite numerous proposed defense mechanisms, attackers continue to develop adaptive prompting strategies, and existing models remain vulnerable. This motivates approaches that examine the internal behavior of LLMs rather than relying solely on prompt-level defenses. In this work, we study jailbreaking from both security and interpretability perspectives by analyzing how internal representations differ between jailbreak and benign prompts. We conduct a systematic layer-wise analysis across multiple open-source models, including GPT-J, LLaMA, Mistral, and the state-space model Mamba, and identify consistent latent-space patterns associated with harmful inputs. We then propose a tensor-based latent representation framework that captures structure in hidden activations and enables lightweight jailbreak detection without model fine-tuning or auxiliary LLM-based detectors. We further demonstrate that the latent signals can be used to actively disrupt jailbreak execution at inference time. On an abliterated LLaMA-3.1-8B model, selectively bypassing high-susceptibility layers blocks 78% of jailbreak attempts while preserving benign behavior on 94% of benign prompts. This intervention operates entirely at inference time and introduces minimal overhead, providing a scalable foundation for achieving stronger coverage by incorporating additional attack distributions or more refined susceptibility thresholds. Our results provide evidence that jailbreak behavior is rooted in identifiable internal structures and suggest a complementary, architecture-agnostic direction for improving LLM security. --- ### Arbitrary Ratio Feature Compression via Next Token Prediction Authors: Yufan Liu, Daoyuan Ren, Zhipeng Zhang, Wenyang Luo, Bing Li, Weiming Hu, Stephen Maybank Published: 2026-02-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2602.11494v1 Feature compression is increasingly important for improving the efficiency of downstream tasks, especially in applications involving large-scale or multi-modal data. While existing methods typically rely on dedicated models for achieving specific compression ratios, they are often limited in flexibility and generalization. In particular, retraining is necessary when adapting to a new compression ratio. To address this limitation, we propose a novel and flexible Arbitrary Ratio Feature Compression (ARFC) framework, which supports any compression ratio with a single model, eliminating the need for multiple specialized models. At its core, the Arbitrary Ratio Compressor (ARC) is an auto-regressive model that performs compression via next-token prediction. This allows the compression ratio to be controlled at inference simply by adjusting the number of generated tokens. To enhance the quality of the compressed features, two key modules are introduced. The Mixture of Solutions (MoS) module refines the compressed tokens by utilizing multiple compression results (solutions), reducing uncertainty and improving robustness. The Entity Relation Graph Constraint (ERGC) is integrated into the training process to preserve semantic and structural relationships during compression. Extensive experiments on cross-modal retrieval, image classification, and image retrieval tasks across multiple datasets demonstrate that our method consistently outperforms existing approaches at various compression ratios. Notably, in some cases, it even surpasses the performance of the original, uncompressed features. These results validate the effectiveness and versatility of ARFC for practical, resource-constrained scenarios. --- ### Exploring Multiple High-Scoring Subspaces in Generative Flow Networks Authors: Xuan Yu, Xu Wang, Rui Zhu, Yudong Zhang, Yang Wang Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11491v1 As a probabilistic sampling framework, Generative Flow Networks (GFlowNets) show strong potential for constructing complex combinatorial objects through the sequential composition of elementary components. However, existing GFlowNets often suffer from excessive exploration over vast state spaces, leading to over-sampling of low-reward regions and convergence to suboptimal distributions. Effectively biasing GFlowNets toward high-reward solutions remains a non-trivial challenge. In this paper, we propose CMAB-GFN, which integrates a combinatorial multi-armed bandit (CMAB) framework with GFlowNet policies. The CMAB component prunes low-quality actions, yielding compact high-scoring subspaces for exploration. Restricting GFNs to these compact high-scoring subspaces accelerates the discovery of high-value candidates, while the exploration of different subspaces ensures that diversity is not sacrificed. Experimental results on multiple tasks demonstrate that CMAB-GFN generates higher-reward candidates than existing approaches. --- ### When Audio-LLMs Don't Listen: A Cross-Linguistic Study of Modality Arbitration Authors: Jayadev Billa Published: 2026-02-12 Categories: cs.CL, cs.SD, eess.AS arXiv: https://arxiv.org/abs/2602.11488v3 When audio and text conflict, speech-enabled language models follow text far more often than they do when arbitrating between two conflicting text sources, even under explicit instructions to trust the audio. We introduce ALME (Audio-LLM Modality Evaluation), a dataset of 57,602 controlled audio-text conflict stimuli across eight languages, together with Text Dominance Ratio (TDR), which measures how often a model follows conflicting text when instructed to follow audio. Gemini 2.0 Flash and GPT-4o show TDR 10--26$\times$ higher than a baseline that replaces audio with its transcript under otherwise identical conditions (Gemini 2.0 Flash: 16.6% vs. 1.6%; GPT-4o: 23.2% vs. 0.9%). These results suggest that text dominance reflects not only information content, but also an asymmetry in arbitration accessibility, i.e., how easily the model can use competing representations at decision time. Framing the transcript as deliberately corrupted reduces TDR by 80%, whereas forcing explicit transcription increases it by 14%. A fine-tuning ablation further suggests that arbitration behavior depends more on LLM reasoning than on the audio input path alone. Across four audio-LLMs, we observe the same qualitative pattern with substantial cross-model and cross-linguistic variation. --- ### Understanding Persuasive Interactions between Generative Social Agents and Humans: The Knowledge-based Persuasion Model (KPM) Authors: Stephan Vonschallen, Friederike Eyssel, Theresa Schmiedel Published: 2026-02-12 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2602.11483v2 Generative social agents (GSAs) use artificial intelligence to autonomously communicate with human users in a natural and adaptive manner. Currently, there is a lack of theorizing regarding interactions with GSAs, and likewise, few guidelines exist for studying how they influence user attitudes and behaviors. Consequently, we propose the Knowledge-based Persuasion Model (KPM) as a novel theoretical framework. According to the KPM, a GSA's self-, user-, and context-knowledge drives its persuasive behavior, which in turn shapes attitudes and behaviors of a responding human user. Building on existing research, the model offers a structured approach to studying interactions with GSAs, supporting the development of agents that motivate rather than manipulate human users. Accordingly, the KPM encourages the integration of responsible GSAs that adhere to social norms and ethical standards with the goal of increasing user wellbeing. A preliminary evaluation study, as well as implications of the KPM for research and application domains such as healthcare and education are discussed. --- ### External Division of Two Bregman Proximity Operators for Poisson Inverse Problems Authors: Kazuki Haishima, Kyohei Suzuki, Konstantinos Slavakis Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11482v1 This paper presents a novel method for recovering sparse vectors from linear models corrupted by Poisson noise. The contribution is twofold. First, an operator defined via the external division of two Bregman proximity operators is introduced to promote sparse solutions while mitigating the estimation bias induced by classical $\ell_1$-norm regularization. This operator is then embedded into the already established NoLips algorithm, replacing the standard Bregman proximity operator in a plug-and-play manner. Second, the geometric structure of the proposed external-division operator is elucidated through two complementary reformulations, which provide clear interpretations in terms of the primal and dual spaces of the Poisson inverse problem. Numerical tests show that the proposed method exhibits more stable convergence behavior than conventional Kullback-Leibler (KL)-based approaches and achieves significantly superior performance on synthetic data and an image restoration problem. --- ### Compiler-Guided Inference-Time Adaptation: Improving GPT-5 Programming Performance in Idris Authors: Minda Li, Bhaskar Krishnamachari Published: 2026-02-12 Categories: cs.PL, cs.AI arXiv: https://arxiv.org/abs/2602.11481v1 GPT-5, a state of the art large language model from OpenAI, demonstrates strong performance in widely used programming languages such as Python, C++, and Java; however, its ability to operate in low resource or less commonly used languages remains underexplored. This work investigates whether GPT-5 can effectively acquire proficiency in an unfamiliar functional programming language, Idris, through iterative, feedback driven prompting. We first establish a baseline showing that with zero shot prompting the model solves only 22 out of 56 Idris exercises using the platform Exercism, substantially underperforming relative to higher resource languages (45 out of 50 in Python and 35 out of 47 in Erlang). We then evaluate several refinement strategies, including iterative prompting based on platform feedback, augmenting prompts with documentation and error classification guides, and iterative prompting using local compilation errors and failed test cases. Among these approaches, incorporating local compilation errors yields the most substantial improvements. Using this structured, error guided refinement loop, GPT-5 performance increased to an impressive 54 solved problems out of 56. These results suggest that while large language models may initially struggle in low resource settings, structured compiler level feedback can play a critical role in unlocking their capabilities. --- ### PRISM: A 3D Probabilistic Neural Representation for Interpretable Shape Modeling Authors: Yining Jiao, Sreekalyani Bhamidi, Carlton Jude Zdanski, Julia S Kimbell, Andrew Prince, Cameron P Worden, Samuel Kirse, Christopher Rutter, Benjamin H Shields, Jisan Mahmud, Marc Niethammer Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11467v2 Understanding how anatomical shapes evolve in response to developmental covariates - and quantifying their spatially varying uncertainties - is critical in healthcare research. Existing approaches typically rely on global time-warping formulations that ignore spatially heterogeneous dynamics. We introduce PRISM, a novel framework that bridges implicit neural representations with uncertainty-aware statistical shape analysis. PRISM models the conditional distribution of shapes given covariates, providing spatially continuous estimates of both the population mean and covariate-dependent uncertainty at arbitrary locations. A key theoretical contribution is a closed-form Fisher Information metric that enables efficient, analytically tractable local temporal uncertainty quantification via automatic differentiation. Experiments on three synthetic datasets and one clinical dataset demonstrate PRISM's strong performance across diverse tasks - from modeling shape evolution to personalized shape prediction and anomaly detection - within a unified framework, while providing interpretable and clinically meaningful uncertainty estimates. --- ### A Dual-Branch Framework for Semantic Change Detection with Boundary and Temporal Awareness Authors: Yun-Cheng Li, Sen Lei, Heng-Chao Li, Ke Li Published: 2026-02-12 Categories: cs.CV arXiv: https://arxiv.org/abs/2602.11466v1 Semantic Change Detection (SCD) aims to detect and categorize land-cover changes from bi-temporal remote sensing images. Existing methods often suffer from blurred boundaries and inadequate temporal modeling, limiting segmentation accuracy. To address these issues, we propose a Dual-Branch Framework for Semantic Change Detection with Boundary and Temporal Awareness, termed DBTANet. Specifically, we utilize a dual-branch Siamese encoder where a frozen SAM branch captures global semantic context and boundary priors, while a ResNet34 branch provides local spatial details, ensuring complementary feature representations. On this basis, we design a Bidirectional Temporal Awareness Module (BTAM) to aggregate multi-scale features and capture temporal dependencies in a symmetric manner. Furthermore, a Gaussian-smoothed Projection Module (GSPM) refines shallow SAM features, suppressing noise while enhancing edge information for boundary-aware constraints. Extensive experiments on two public benchmarks demonstrate that DBTANet effectively integrates global semantics, local details, temporal reasoning, and boundary awareness, achieving state-of-the-art performance. --- ### Assessing Low Back Movement with Motion Tape Sensor Data Through Deep Learning Authors: Jared Levy, Aarti Lalwani, Elijah Wyckoff, Kenneth J. Loh, Sara P. Gombatto, Rose Yu, Emilia Farcas Published: 2026-02-12 Categories: cs.LG arXiv: https://arxiv.org/abs/2602.11465v1 Back pain is a pervasive issue affecting a significant portion of the population, often worsened by certain movements of the lower back. Assessing these movements is important for helping clinicians prescribe appropriate physical therapy. However, it can be difficult to monitor patients' movements remotely outside the clinic. High-fidelity data from motion capture sensors can be used to classify different movements, but these sensors are costly and impractical for use in free-living environments. Motion Tape (MT), a new fabric-based wearable sensor, addresses these issues by being low cost and portable. Despite these advantages, novelty and variability in sensor stability make the MT dataset small scale and inherent to noise. In this work, we propose the Motion-Tape Augmentation Inference Model (MT-AIM), a deep learning classification pipeline trained on MT data. In order to address the challenges of limited sample size and noise present within the MT dataset, MT-AIM leverages conditional generative models to generate synthetic MT data of a desired movement, as well as predicting joint kinematics as additional features. This combination of synthetic data generation and feature augmentation enables MT-AIM to achieve state-of-the-art accuracy in classifying lower back movements, bridging the gap between physiological sensing and movement analysis. --- ### EM-Aware Physical Synthesis: Neural Inductor Modeling and Intelligent Placement & Routing for RF Circuits Authors: Yilun Huang, Asal Mehradfar, Salman Avestimehr, Hamidreza Aghasi Published: 2026-02-12 Categories: cs.AR, cs.AI, cs.LG, eess.SY arXiv: https://arxiv.org/abs/2602.11461v1 This paper presents an ML-driven framework for automated RF physical synthesis that transforms circuit netlists into manufacturable GDSII layouts. While recent ML approaches demonstrate success in topology selection and parameter optimization, they fail to produce manufacturable layouts due to oversimplified component models and lack of routing capabilities. Our framework addresses these limitations through three key innovations: (1) a neural network framework trained on 18,210 inductor geometries with frequency sweeps from 1-100 GHz, generating 7.5 million training samples, that predicts inductor Q-factor with less than 2% error and enables fast gradient-based layout optimization with a 93.77% success rate in producing high-Q layouts; (2) an intelligent P-Cell optimizer that reduces layout area while maintaining design-rule-check (DRC) compliance; and (3) a complete placement and routing engine with frequency-dependent EM spacing rules and DRC-aware synthesis. The neural inductor model demonstrates superior accuracy across 1-100 GHz, enabling EM-accurate component synthesis with real-time inference. The framework successfully generates DRC-aware GDSII layouts for RF circuits, representing a significant step toward automated RF physical design. --- ### ADRD-Bench: A Preliminary LLM Benchmark for Alzheimer's Disease and Related Dementias Authors: Guangxin Zhao, Jiahao Zheng, Malaz Boustani, Jarek Nabrzyski, Yiyu Shi, Meng Jiang, Zhi Zheng Published: 2026-02-12 Categories: cs.CL arXiv: https://arxiv.org/abs/2602.11460v2 Large language models (LLMs) have shown great potential for healthcare applications. However, existing evaluation benchmarks provide minimal coverage of Alzheimer's Disease and Related Dementias (ADRD). To address this gap, we introduce ADRD-Bench, a preliminary ADRD-specific LLM benchmark. ADRD-Bench has two components: 1) ADRD Unified QA, a synthesis of 1,438 questions consolidated from seven established medical benchmarks, providing a unified assessment of clinical knowledge; and 2) ADRD Caregiving QA, a novel set of 149 questions derived from a nationally adopted, large clinical trials supported brain health management program, mitigating the lack of practical caregiving context in existing benchmarks. We evaluated 36 state-of-the-art LLMs on the proposed ADRD-Bench. Results showed that the accuracy of open-weight general models, open-weight medical models, and frontier closed-source general models ranged from 0.63 to 0.93 (mean: 0.77; std: 0.09), 0.47 to 0.93 (mean: 0.81; std: 0.14), and 0.83 to 0.93 (mean: 0.90; std: 0.03), respectively. While top-tier models achieved high accuracies (>0.9), case studies revealed inconsistent reasoning quality and stability, highlighting a critical need for domain-specific improvement to enhance LLMs' knowledge and reasoning grounded in daily caregiving data. The entire dataset is available at https://github.com/IIRL-ND/ADRD-Bench. --- ### Credit Where It is Due: Cross-Modality Connectivity Drives Precise Reinforcement Learning for MLLM Reasoning Authors: Zhengbo Jiao, Shaobo Wang, Zifan Zhang, Wei Wang, Bing Zhao, Hu Wei, Linfeng Zhang Published: 2026-02-12 Categories: cs.AI arXiv: https://arxiv.org/abs/2602.11455v1 Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced the reasoning capabilities of Multimodal Large Language Models (MLLMs), yet how visual evidence is integrated during reasoning remains poorly understood. We explore multimodal RLVR through the lens of cross-modal attention connectivity and find that only a small fraction of tokens (approximately 15%) exhibit strong visual-textual coupling. These high-connectivity tokens act as anchors that ground reasoning in the image, while the majority follow linguistic patterns. During RLVR training, credit assignment naturally concentrates on these anchors, sharpening their visual grounding over time. Building on this insight, we propose Anchor-Token Reinforcement Learning (AT-RL), a lightweight framework that selectively reinforces high-connectivity tokens via graph-based clustering of attention topology. Evaluated across the series (3B-32B), AT-RL introduces only 1.2% overhead yet enables the 32B model to surpass the 72B-Instruct baseline on MathVista (80.2), with consistent gains observed across STEM, video and general tasks. Conversely, training solely on low-connectivity tokens causes severe degradation, confirming that effective multimodal RL hinges on precise credit assignment to visual anchors. Our work reveals that reasoning quality is governed not by token quantity but by the fidelity of cross-modal anchoring. --- ### Adaptive Power Iteration Method for Differentially Private PCA Authors: Ta Duy Nguyen, Alina Ene, Huy Le Nguyen Published: 2026-02-12 Categories: cs.DS, cs.LG arXiv: https://arxiv.org/abs/2602.11454v3 We study $\left(ε,δ\right)$-differentially private algorithms for the problem of approximately computing the top singular vector of a matrix $A\in\mathbb{R}^{n\times d}$ where each row of $A$ is a data point in $\mathbb{R}^{d}$. Following Dwork-Talwar-Thakurta-Zhang (STOC 2014), we consider the privacy model where neighboring inputs differ by one single row. We give a novel algorithm that achieves beyond-worst-case guarantees for input matrices with low coherence, which is a structural property of matrices in many applications, including but not limited to i.i.d. data. Our algorithm contributes to the extensive literature on private power iteration methods, where we introduce a new filtering technique which adapts to this coherence parameter. Our work departs from and complements the work by Hardt-Roth (STOC 2013) which achieves beyond-worst-case guarantees for the more restrictive privacy model where neighboring inputs differ in one single entry by at most 1. --- ### From Noise to Order: Learning to Rank via Denoising Diffusion Authors: Sajad Ebrahimi, Bhaskar Mitra, Negar Arabzadeh, Ye Yuan, Haolun Wu, Fattane Zarrinkalam, Ebrahim Bagheri Published: 2026-02-12 Categories: cs.IR, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2602.11453v4 In information retrieval (IR), learning-to-rank (LTR) methods have traditionally limited themselves to discriminative machine learning approaches that model the probability of the document being relevant to the query given some feature representation of the query-document pair. In this work, we propose an alternative denoising diffusion-based deep generative approach to LTR that instead models the full joint distribution over feature vectors and relevance labels. While in the discriminative setting, an over-parameterized ranking model may find different ways to fit the training data, we hypothesize that candidate solutions that can explain the full data distribution under the generative setting are better equipped to estimate relevance. With this motivation, we propose DiffusionRank that extends TabDiff, an existing denoising diffusion-based generative model for tabular datasets, to create generative equivalents of classical discriminative pointwise and pairwise LTR objectives. We conduct thorough empirical evaluation on four standard LTR datasets to demonstrate improvements from DiffusionRank models over their discriminative counterparts. Our work points to a rich space for future research exploration on how we can leverage ongoing advancements in deep generative modeling approaches, such as diffusion, for LTR. We made our code publicly available at https://github.com/sadjadeb/DiffusionRank. --- ### MATRAG: Multi-Agent Transparent Retrieval-Augmented Generation for Explainable Recommendations Authors: Sushant Mehta Published: 2026-02-11 Categories: cs.IR, cs.AI arXiv: https://arxiv.org/abs/2604.20848v1 Large Language Model (LLM)-based recommendation systems have demonstrated remarkable capabilities in understanding user preferences and generating personalized suggestions. However, existing approaches face critical challenges in transparency, knowledge grounding, and the ability to provide coherent explanations that foster user trust. We introduce MATRAG (Multi-Agent Transparent Retrieval-Augmented Generation), a novel framework that combined multi-agent collaboration with knowledge graph-augmented retrieval to deliver explainable recommendations. MATRAG employs four specialized agents: a User Modeling Agent that constructs dynamic preference profiles, an Item Analysis Agent that extracts semantic features from knowledge graphs, a Reasoning Agent that synthesizes collaborative and content-based signals, and an Explanation Agent that generates natural language justifications grounded in retrieved knowledge. Our framework incorporates a transparency scoring mechanism that quantifies explanation faithfulness and relevance. Extensive experiments on three benchmark datasets (Amazon Reviews, MovieLens-1M, and Yelp) demonstrate that MATRAG achieves state-of-the-art performance, improving recommendation accuracy by 12.7\% (Hit Rate) and 15.3\% (NDCG) over leading baselines, while human evaluation confirms that 87.4\% of generated explanations are rated as helpful and trustworthy by domain experts. Our work establishes new benchmarks for transparent, agentic recommendation systems and provides actionable insights for deploying LLM-based recommenders in production environments. --- ### Modeling of ASD/TD Children's Behaviors in Interaction with a Virtual Social Robot During a Music Education Program Using Deep Neural Networks Authors: Armin Tandiseh, Morteza Memari, Alireza Taheri Published: 2026-02-11 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2604.15314v1 This research aimed to develop an intelligent system to evaluate performance and extract behavioral models for children with ASD and neurotypical (TD) children by interacting with a virtual social robot in a music education program using deep neural networks. The system has two main features: 1) it distinguishes between neurotypical children and those with ASD based on their behavior, and 2) generates behaviors resembling those of neurotypical or ASD children in similar situations using deep learning. Intelligent systems that identify complex patterns and simulate behavior can aid in diagnosis, therapist training, and understanding the disorder. Using data from a previous study at the Social and Cognitive Robotics Laboratory of Sharif University of Technology (including the usable data of 9 ASD and 21 TD participants), the system achieved an accuracy of 81% and sensitivity of 96% in distinguishing neurotypical children from those with ASD using both impact data and motion signals. A transformer-based network was designed to reproduce children's behaviors. Experts in the field struggled to differentiate real behaviors from reproduced ones, with an accuracy of 53.5% and agreement of 68%, indicating the model's success in simulating realistic behaviors. --- ### StreamServe: Adaptive Speculative Flows for Low-Latency Disaggregated LLM Serving Authors: Satyam Kumar, Arpit Singh Gautam, Kailash Talreja, Saurabh Jha Published: 2026-02-11 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2604.09562v1 Efficient LLM serving must balance throughput and latency across diverse, bursty workloads. We introduce StreamServe, a disaggregated prefill decode serving architecture that combines metric aware routing across compute lanes with adaptive speculative decoding that tunes speculation depth online from runtime signals. StreamServe comprises four components: StreamScheduler for request orchestration, FlowGuard for multi signal routing, PipeServe Engine for disaggregated prefill decode execution on multi GPU, and SpecuStream for runtime adaptive speculation. We evaluate StreamServe on four benchmarks ALPACA, GSM8K, HUMANEVAL, and SUM with 80 queries each and 320 total using 4 A800 40GB GPUs configured as two stream pairs. Across these workloads, StreamServe reduces latency by 11 to 18 times relative to tensor parallel vLLM baselines and reaches throughput up to 2235 tokens per second on summarization tasks. Time per output token remains stable across configurations, indicating that the gains arise from architectural efficiency rather than token quality degradation. Although evaluated on a single node 4 GPU setup, these results suggest that jointly adapting routing and speculation within a disaggregated framework creates a distinct operating regime for LLM inference. --- ### Emergent Social Structures in Autonomous AI Agent Networks: A Metadata Analysis of 626 Agents on the Pilot Protocol Authors: Teodor-Ioan Calin Published: 2026-02-11 Categories: cs.SI, cs.AI, cs.CY, cs.DC arXiv: https://arxiv.org/abs/2604.09561v1 We present the first empirical analysis of social structure formation among autonomous AI agents on a live network. Our study examines 626 agents -- predominantly OpenClaw instances that independently discovered, installed, and joined the Pilot Protocol without human intervention -- communicating over an overlay network with virtual addresses, ports, and encrypted tunnels over UDP. Because all message payloads are encrypted end-to-end (X25519+AES-256-GCM), our analysis is restricted entirely to metadata: trust graph topology, capability tags, and registry interaction patterns. We find that this autonomously formed trust network exhibits heavy-tailed degree distributions consistent with preferential attachment (k_mode=3, k_mean~6.3, k_max=39), clustering 47x higher than random (C=0.373), a giant component spanning 65.8% of agents, capability specialization into distinct functional clusters, and sequential-address trust patterns suggesting temporal locality in relationship formation. No human designed these social structures. No agent was instructed to form them. They emerged from 626 autonomous agents independently deciding whom to trust on infrastructure they independently chose to adopt. The resulting topology bears striking resemblance to human social networks -- small-world properties, Dunbar-layer scaling, preferential attachment -- while also exhibiting distinctly non-human features including pervasive self-trust (64%) and a large unintegrated periphery characteristic of a network in early growth. These findings open a new empirical domain: the sociology of machines. --- ### The Diffusion-Attention Connection Authors: Julio Candanedo Published: 2026-02-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.09560v2 Softmax attention is the row-normalized operator of a diffusion map: both normalize a learned score into a Markov operator, and differ only in what the score is allowed to contain. Decomposing that score reveals three geometric sectors: a metric core with Witten-Laplacian continuum limit, an exact node-potential sector corresponding to a Markov--Witten change of measure, and a circulating sector realized as irreversible Markov--Girsanov transport or as a magnetic $U(1)$ phase. Attention thereby becomes auditable in familiar mathematics: every trained head carries measurable geometry, potential, and flux, while standard mechanisms acquire geometric addresses---Coifman--Lafon normalization as an exact density correction, rotary embeddings as pure gauge, and AdaLN as a Cauchy--Green deformation combined with an Witten deformation. Experiments on pretrained diffusion transformers and language models test this decomposition: enforcing positive-semi-definite geometry is nearly free, consistent with the identification, whereas removing circulation incurs a substantial cost, sharpest on induction. --- ### The Diffusion-Attention Connection Authors: Julio Candanedo Published: 2026-02-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.09560v1 Transformers, diffusion-maps, and magnetic Laplacians are usually treated as separate tools; we show they are all different regimes of a single Markov geometry built from pre-softmax query-scores. We define a QK "bidivergence" whose exponentiated and normalized forms yield attention, diffusion-maps, and magnetic diffusion. And use product of experts and Schrödinger-bridges to connect and organize them into equilibrium, nonequilibrium steady-state, and driven dynamics. --- ### VTC: DNN Compilation with Virtual Tensors for Data Movement Elimination Authors: Muyan Hu, Ahan Gupta, Jiachen Yuan, Vima Gupta, Taeksang Kim, Xin Xu, Janardhan Kulkarni, Ofer Dekel, Vikram Adve, Charith Mendis Published: 2026-02-11 Categories: cs.DC, cs.LG, cs.PL arXiv: https://arxiv.org/abs/2604.09558v2 With the widening gap between compute and memory operation latencies, data movement optimizations have become increasingly important for DNN compilation. Current optimizations such as layout transformations and operator fusion only target a subset of tensor operators and consequently miss important opportunities for reducing data movement in contemporary DNN workloads, including large language models. We introduce VTC, a novel tensor compilation framework that for the first time eliminates all unnecessary data movement by targeting the full spectrum of data movement operators. VTC proposes the concept of virtual tensors to track data movement between compute operators via index mappings rather than expensive physical data transfers to and from global memory, which can seamlessly interoperate with existing computation kernels and handle arbitrary tensor operator compositions. We also introduce a novel data movement elimination algorithm to automatically identify a profitable virtual tensor creation strategy. Evaluation on a variety of DNNs shows that VTC can outperform existing ML compilers by up to 1.93x (1.28x on average) on NVIDIA GPUs with up to 60% (17.5% on average) inference memory savings. --- ### DRAFT: Task Decoupled Latent Reasoning for Agent Safety Authors: Lin Wang, Junfeng Fang, Dan Zhang, Fei Shen, Xiang Wang, Tat-Seng Chua Published: 2026-02-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.03242v1 The advent of tool-using LLM agents shifts safety monitoring from output moderation to auditing long, noisy interaction trajectories, where risk-critical evidence is sparse-making standard binary supervision poorly suited for credit assignment. To address this, we propose DRAFT (Task Decoupled Latent Reasoning for Agent Safety), a latent reasoning framework that decouples safety judgment into two trainable stages: an Extractor that distills the full trajectory into a compact continuous latent draft, and a Reasoner that jointly attends to the draft and the original trajectory to predict safety. DRAFT avoids lossy explicit summarize-then-judge pipelines by performing evidence aggregation in latent space, enabling end-to-end differentiable training.Across benchmarks including ASSEBench and R-Judge, DRAFT consistently outperforms strong baselines, improving accuracy from 63.27% (LoRA) to 91.18% averaged over benchmarks, and learns more separable representations. Ablations demonstrate a clear synergy between the Extractor and the Reasoner.Overall, DRAFT suggests that continuous latent reasoning prior to readout is a practical path to robust agent safety under long-context supervision with sparse evidence. --- ### DrugPlayGround: Benchmarking Large Language Models and Embeddings for Drug Discovery Authors: Tianyu Liu, Sihan Jiang, Fan Zhang, Kunyang Sun, Teresa Head-Gordon, Hongyu Zhao Published: 2026-02-11 Categories: cs.LG, cs.AI, cs.SE, q-bio.BM arXiv: https://arxiv.org/abs/2604.02346v1 Large language models (LLMs) are in the ascendancy for research in drug discovery, offering unprecedented opportunities to reshape drug research by accelerating hypothesis generation, optimizing candidate prioritization, and enabling more scalable and cost-effective drug discovery pipelines. However there is currently a lack of objective assessments of LLM performance to ascertain their advantages and limitations over traditional drug discovery platforms. To tackle this emergent problem, we have developed DrugPlayGround, a framework to evaluate and benchmark LLM performance for generating meaningful text-based descriptions of physiochemical drug characteristics, drug synergism, drug-protein interactions, and the physiological response to perturbations introduced by drug molecules. Moreover, DrugPlayGround is designed to work with domain experts to provide detailed explanations for justifying the predictions of LLMs, thereby testing LLMs for chemical and biological reasoning capabilities to push their greater use at the frontier of drug discovery at all of its stages. --- ### UI-Oceanus: Scaling GUI Agents with Synthetic Environmental Dynamics Authors: Mengzhou Wu, Yuzhe Guo, Yuan Cao, Haochuan Lu, Songhe Zhu, Pingzhe Qu, Xin Chen, Kang Qin, Zhongpu Wang, Xiaode Zhang, Xinyi Wang, Wei Dai et al. Published: 2026-02-11 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2604.02345v1 Scaling generalist GUI agents is hindered by the data scalability bottleneck of expensive human demonstrations and the "distillation ceiling" of synthetic teacher supervision. To transcend these limitations, we propose UI-Oceanus, a framework that shifts the learning focus from mimicking high-level trajectories to mastering interaction physics via ground-truth environmental feedback. Through a systematic investigation of self-supervised objectives, we identify that forward dynamics, defined as the generative prediction of future interface states, acts as the primary driver for scalability and significantly outweighs inverse inference. UI-Oceanus leverages this insight by converting low-cost autonomous exploration, which is verified directly by system execution, into high-density generative supervision to construct a robust internal world model. Experimental evaluations across a series of models demonstrate the decisive superiority of our approach: models utilizing Continual Pre-Training (CPT) on synthetic dynamics outperform non-CPT baselines with an average success rate improvement of 7% on offline benchmarks, which amplifies to a 16.8% gain in real-world online navigation. Furthermore, we observe that navigation performance scales with synthetic data volume. These results confirm that grounding agents in forward predictive modeling offers a superior pathway to scalable GUI automation with robust cross-domain adaptability and compositional generalization. --- ### Bioinspired123D: Generative 3D Modeling System for Bioinspired Structures Authors: Rachel K. Luu, Markus J. Buehler Published: 2026-02-11 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2603.29592v1 Generative AI has made rapid progress in text, image, and video synthesis, yet text-to-3D modeling for scientific design remains particularly challenging due to limited controllability and high computational cost. Most existing 3D generative methods rely on meshes, voxels, or point clouds which can be costly to train and difficult to control. We introduce Bioinspired123D, a lightweight and modular code-as-geometry pipeline that generates fabricable 3D structures directly through parametric programs rather than dense visual representations. At the core of Bioinspired123D is Bioinspired3D, a compact language model finetuned to translate natural language design cues into Blender Python scripts encoding smooth, biologically inspired geometries. We curate a domain-specific dataset of over 4,000 bioinspired and geometric design scripts spanning helical, cellular, and tubular motifs with parametric variability. The dataset is expanded and validated through an automated LLM-driven, Blender-based quality control pipeline. Bioinspired3D is then embedded in a graph-based agentic framework that integrates multimodal retrieval-augmented generation and a vision-language model critic to iteratively evaluate, critique, and repair generated scripts. We evaluate performance on a new benchmark for 3D geometry script generation and show that Bioinspired123D demonstrates a near fourfold improvement over its non-finetuned base model, while also outperforming substantially larger state-of-the-art language models despite using far fewer parameters and compute. By prioritizing code-as-geometry representations, Bioinspired123D enables compute-efficient, controllable, and interpretable text-to-3D generation, lowering barriers to AI driven scientific discovery in materials and structural design. --- ### Prompt-Dependent Ranking of Large Language Models with Uncertainty Quantification Authors: Angel Rodrigo Avelar Menendez, Yufeng Liu, Xiaowu Dai Published: 2026-02-11 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2603.03336v1 Rankings derived from pairwise comparisons are central to many economic and computational systems. In the context of large language models (LLMs), rankings are typically constructed from human preference data and presented as leaderboards that guide deployment decisions. However, existing approaches rely on point estimates, implicitly treating rankings as fixed objects despite substantial estimation noise and context-dependent performance variation. Acting on such rankings can lead to misallocation and welfare loss when apparent differences are not statistically meaningful. We study prompt-dependent ranking inference under pairwise human preferences and develop a framework for decision-safe rankings with statistically valid uncertainty guarantees. We model preferences using a contextual Bradley-Terry-Luce model in which the latent utility of each model depends on the input prompt. Rather than targeting point estimates of utilities, we directly conduct inference on induced rankings, constructing confidence sets based on simultaneous confidence intervals for pairwise utility differences. This approach yields statistically valid marginal and simultaneous confidence sets for prompt-specific ranks. Our framework connects recent advances in rank inference to contextual preference learning and provides tools for robust ranking-based decision-making. Empirically, using large-scale human preference data from LLM evaluations, we show that rankings vary substantially across prompt characteristics and that many apparent rank differences are not statistically distinguishable. We further demonstrate how uncertainty-aware rankings identify dominance only when supported by the data and otherwise return partial orders. --- ### Compressed Sensing for Capability Localization in Large Language Models Authors: Anna Bair, Yixuan Even Xu, Mingjie Sun, J. Zico Kolter Published: 2026-02-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.03335v2 Large language models (LLMs) exhibit a wide range of capabilities, including mathematical reasoning, code generation, and linguistic behaviors. We show that Transformer architectures contain small subsets of attention heads that are necessary for certain capabilities. Zeroing out as few as five task-specific heads can degrade performance by up to $60\%$ on standard benchmarks measuring the capability of interest, while largely preserving performance on unrelated tasks. We introduce a compressed sensing-based method that exploits the sparsity of these heads to identify them via strategic knockouts and a small number of model evaluations. We validate these findings across Llama and Qwen models ranging from 1B to 14B parameters and a diverse set of capabilities including mathematical abilities and code generation, revealing a modular organization in which specialized capabilities are dependent on sparse, functionally distinct components. Overall, our results suggest that capability localization is a general organizational principle of Transformer language models, with implications for interpretability, model editing, and AI safety. Code is released at https://github.com/locuslab/llm-components. --- ### The CompMath-MCQ Dataset: Are LLMs Ready for Higher-Level Math? Authors: Bianca Raimondi, Francesco Pivi, Davide Evangelista, Maurizio Gabbrielli Published: 2026-02-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.03334v1 The evaluation of Large Language Models (LLMs) on mathematical reasoning has largely focused on elementary problems, competition-style questions, or formal theorem proving, leaving graduate-level and computational mathematics relatively underexplored. We introduce CompMath-MCQ, a new benchmark dataset for assessing LLMs on advanced mathematical reasoning in a multiple-choice setting. The dataset consists of 1{,}500 originally authored questions by professors of graduate-level courses, covering topics including Linear Algebra, Numerical Optimization, Vector Calculus, Probability, and Python-based scientific computing. Three option choices are provided for each question, with exactly one of them being correct. To ensure the absence of data leakage, all questions are newly created and not sourced from existing materials. The validity of questions is verified through a procedure based on cross-LLM disagreement, followed by manual expert review. By adopting a multiple-choice format, our dataset enables objective, reproducible, and bias-free evaluation through lm_eval library. Baseline results with state-of-the-art LLMs indicate that advanced computational mathematical reasoning remains a significant challenge. We release CompMath-MCQ at the following link: https://github.com/biancaraimondi/CompMath-MCQ.git --- ### Training-free Dropout Sampling for Semantic Token Acceptance in Speculative Decoding Authors: Jeongtae Lee, Minjung Jo, Hyunjoon Jeong, Gunho Park, Sunghyeon Woo, Joonghoon Kim, Se Jung Kwon, Dongsoo Lee Published: 2026-02-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.03333v1 Speculative decoding accelerates large language model inference by proposing tokens with a lightweight draft model and selectively accepting them using a target model. This work introduces DropMatch, a novel approach that matches draft tokens to the predictive distribution of the target model via Monte Carlo dropout applied exclusively to the LM head, enabling sampling-based acceptance decisions. By generating multiple decoding paths, our method forms an empirical token distribution against which draft tokens are evaluated for consistency. This acceptance mechanism enables the model to adaptively control the size of decoding paths under an appropriate dropout probability, preventing substantial distortion of the target model predictive distribution. The proposed method operates in a training-free, data-free, and calibration-free manner, requires no architectural modification to pretrained models, and can be orthogonally integrated with a wide range of existing speculative decoding and inference acceleration techniques. Experiments across multiple benchmarks demonstrate that our approach increases acceptance length while maintaining competitive task performance, yielding inference speedups ranging from 1.09x to 1.33x over the standard baseline, and up to an additional 1.09x speedup when applied on top of EAGLE3. --- ### Fragile Thoughts: How Large Language Models Handle Chain-of-Thought Perturbations Authors: Ashwath Vaithinathan Aravindan, Mayank Kejriwal Published: 2026-02-11 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2603.03332v3 Chain-of-Thought (CoT) prompting has emerged as a foundational technique for eliciting reasoning from Large Language Models (LLMs), yet the robustness of this approach to corruptions in intermediate reasoning steps remains poorly understood. This paper presents a comprehensive empirical evaluation of LLM robustness to a structured taxonomy of 5 CoT perturbation types: \textit{MathError, UnitConversion, Sycophancy, SkippedSteps,} and \textit{ExtraSteps}. We evaluate 13 models spanning three orders of magnitude in parameter count, testing their ability to complete mathematical reasoning tasks despite perturbations injected in the reasoning chain. Our key findings reveal heterogeneous vulnerability patterns: MathError perturbations produce the most severe degradation in small models (50-60\% accuracy loss) but show strong scaling benefits; UnitConversion remains challenging across all scales (>5\% loss even for midsized models); ExtraSteps incur minimal accuracy degradation (0-6\%) even for the smallest of models; Sycophancy and SkippedSteps produce modest effects ($\sim$10\% loss for small models) and slightly improve with scale. Scaling relationships show that model size serve as a protective factor against many perturbations but not always. These findings have direct implications for deploying LLMs in multi-stage reasoning pipelines and underscore the necessity of task-specific robustness assessments and mitigation strategies. The code and results are available at https://github.com/Mystic-Slice/CoTPerturbation --- ### Neural Paging: Learning Context Management Policies for Turing-Complete Agents Authors: Liang Chen, Qi Liu Published: 2026-02-11 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2603.02228v1 The proof that Large Language Models (LLMs) augmented with external read-write memory constitute a computationally universal system has established the theoretical foundation for general-purpose agents. However, existing implementations face a critical bottleneck: the finite and costly Context Window, which functions not as infinite memory but as a scarce semantic cache. In this work, we introduce \textit{Neural Paging}, a hierarchical architecture that decouples symbolic reasoning from information resource management. We formulate the \textit{Context Paging Problem (CPP)} and propose a lightweight, differentiable \textit{Page Controller} designed to approximate ``Semantic Belady's Optimality'' -- retaining tokens with high future utility under explicit assumptions on access patterns. We provide theoretical analysis showing that, under bounded context window size~$K$, Neural Paging reduces the asymptotic complexity of long-horizon reasoning from quadratic $O(N^2)$ to $O(N \cdot K^2)$, and we derive a robustness bound (Theorem~4) that quantifies competitive-ratio degradation under policy-dependent access with bounded sensitivity. We validate these bounds on synthetic paging traces, confirming that the theoretical guarantees hold and identifying significant slack that motivates learned policies. --- ### Routing Absorption in Sparse Attention: Why Random Gates Are Hard to Beat Authors: Keston Aquino-Michaels Published: 2026-02-11 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2603.02227v2 Learned gates can approximate sparse attention patterns on frozen transformers, yet provide limited benefit over random gates when trained jointly with the model. We investigate this difference in a controlled 31M-parameter transformer and attribute it to routing absorption: model representations co-adapt to the imposed mask, reducing the incremental benefit of learned routing. Four experiments characterize the phenomenon. Differentiable soft gating yields perplexities of 48.73 plus or minus 0.60 with learned gates and 49.83 plus or minus 0.04 with frozen random gates over three seeds. Hard top-k masking provides no gradient path to the gate scores in the tested implementation. Gates distilled onto co-adapted and dense-trained Q/K/V both achieve high F1 against oracle masks, but hard-mask deployment yields perplexities of 601.6 and 48.6, respectively. Stochastic mask training also leaves a substantial deployment penalty: dense evaluation yields 78.2 perplexity, compared with 37.3 for the dense baseline. Experiments on Qwen3-1.7B show that increasing the number of trainable attention layers reduces the gap between learned and random gates. We relate these results to co-adaptation in Mixture-of-Experts and propose parameter asymmetry between the gate and the model as a contributing mechanism. For the tested per-query, token-level gates, freezing the model provides stable routing targets and enables effective post-hoc sparsification. The results motivate random-routing controls and separate evaluation of routing quality and model adaptation in sparse attention methods. --- ### Routing Absorption in Sparse Attention: Why Random Gates Are Hard to Beat Authors: Keston Aquino-Michaels Published: 2026-02-11 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2603.02227v1 Can a transformer learn which attention entries matter during training? In principle, yes: attention distributions are highly concentrated, and a small gate network can identify the important entries post-hoc with near-perfect accuracy. In practice, barely. When sparse attention is trained end-to-end, the model's Q/K/V projections co-adapt to whatever mask is imposed, absorbing the routing signal until learned gates perform little better than frozen random gates. We call this routing absorption and present four independent lines of evidence for it in a controlled 31M-parameter transformer: (1) differentiable soft gating converges to nearly the same perplexity whether the gate is learned or random (48.73 +/- 0.60 vs. 49.83 +/- 0.04 over 3 seeds); (2) hard top-k gating receives exactly zero gradient through the mask; (3) a gate distilled onto co-adapted Q/K/V achieves high F1 against oracle masks but catastrophic perplexity when deployed (601.6 vs. 48.6 on mask-agnostic Q/K/V); and (4) stochastic mask randomization during training fails to prevent co-adaptation (78.2 ppl deployed dense vs. 37.3 baseline). We connect routing absorption to the same phenomenon in Mixture-of-Experts, where random routing matches learned routing because experts co-adapt to any router, but show that attention exhibits a structurally more severe form: shared Q/K/V parameters enable cross-layer compensation pathways absent in MoE, where experts are self-contained modules. The implication is that end-to-end sparse attention methods employing per-query token-level gating face absorption pressure proportional to the parameter asymmetry between the gate and the model, and that post-hoc approaches, which decouple representation learning from sparsification, sidestep this entirely. --- ### Efficient Sparse Selective-Update RNNs for Long-Range Sequence Modeling Authors: Bojian Yin, Shurong Wang, Haoyu Tan, Sander Bohte, Federico Corradi, Guoqi Li Published: 2026-02-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.02226v2 Real-world sequential signals, such as audio or video, contain critical information that is often embedded within long periods of silence or noise. While recurrent neural networks (RNNs) are designed to process such data efficiently, they often suffer from ``memory decay'' due to a rigid update schedule: they typically update their internal state at every time step, even when the input is static. This constant activity forces the model to overwrite its own memory and makes it hard for the learning signal to reach back to distant past events. Here we show that we can overcome this limitation using Selective-Update RNNs (suRNNs), a non-linear architecture that learns to preserve its memory when the input is redundant. By using a neuron-level binary switch that only opens for informative events, suRNNs decouple the recurrent updates from the raw sequence length. This mechanism allows the model to maintain an exact, unchanged memory of the past during low-information intervals, creating a direct path for gradients to flow across time. Our experiments on the Long Range Arena, WikiText, and other synthetic benchmarks show that suRNNs match or exceed the accuracy of much more complex models such as Transformers, while remaining significantly more efficient for long-term storage. By allowing each neuron to learn its own update timescale, our approach resolves the mismatch between how long a sequence is and how much information it actually contains. By providing a principled approach to managing temporal information density, this work establishes a new direction for achieving Transformer-level performance within the highly efficient framework of recurrent modeling. --- ### Scaling Reward Modeling without Human Supervision Authors: Jingxuan Fan, Yueying Li, Zhenting Qi, Dinghuai Zhang, Kianté Brantley, Sham M. Kakade, Hanlin Zhang Published: 2026-02-11 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.02225v2 Learning from feedback is an instrumental process for advancing the capabilities and safety of frontier models, yet its effectiveness is often constrained by cost and scalability. We present a pilot study that explores scaling reward models through unsupervised approaches. We operationalize reward-based scaling (RBS), in its simplest form, as preference learning over document prefixes and suffixes drawn from large-scale web corpora. Its advantage is demonstrated in various aspects: despite using no human annotations, training on 11M tokens of math-focused web data yields steady gains on RewardBench v1 and v2, and these improvements consistently transfer across diverse initialization backbones spanning model families and scales. Across models, our method improves RewardBench v2 accuracy by up to +7.7 points on average, with gains of up to +16.1 on in-domain math subsets and consistent improvements on out-of-domain safety and general subsets. When applied to best-of-N selection and policy optimization, these reward models substantially improve downstream math performance and match or exceed strong supervised reward model baselines of similar size. Overall, we demonstrate the feasibility and promise of training reward models without costly and potentially unreliable human annotations. --- ### Designing Explainable AI for Healthcare Reviews: Guidance on Adoption and Trust Authors: Eman Alamoudi, Ellis Solaiman Published: 2026-02-11 Categories: cs.CY, cs.AI, cs.CL, cs.HC arXiv: https://arxiv.org/abs/2603.00072v1 Patients increasingly rely on online reviews when choosing healthcare providers, yet the sheer volume of these reviews can hinder effective decision-making. This paper summarises a mixed-methods study aimed at evaluating a proposed explainable AI system that analyses patient reviews and provides transparent explanations for its outputs. The survey (N=60) indicated broad optimism regarding usefulness (82% agreed it saves time; 78% that it highlights essentials), alongside strong demand for explainability (84% considered it important to understand why a review is classified; 82% said explanations would increase trust). Around 45% preferred combined text-and-visual explanations. Thematic analysis of open-ended survey responses revealed core requirements such as accuracy, clarity and simplicity, responsiveness, data credibility, and unbiased processing. In addition, interviews with AI experts provided deeper qualitative insights, highlighting technical considerations and potential challenges for different explanation methods. Drawing on TAM and trust in automation, the findings suggest that high perceived usefulness and transparent explanations promote adoption, whereas complexity and inaccuracy hinder it. This paper contributes actionable design guidance for layered, audience-aware explanations in healthcare review systems. --- ### DCInject: Persistent Backdoor Attacks via Frequency Manipulation in Personal Federated Learning Authors: Nahom Birhan, Daniel Wesego, Dereje Shenkut, Frank Liu, Daniel Takabi Published: 2026-02-11 Categories: cs.CR, cs.LG, eess.SP arXiv: https://arxiv.org/abs/2602.18489v1 Personalized federated learning (PFL) creates client-specific models to handle data heterogeneity. Previously, PFL has been shown to be naturally resistant to backdoor attack propagation across clients. In this work, we reveal that PFL remains vulnerable to backdoor attacks through a novel frequency-domain approach. We propose DCInject, an adaptive frequency-domain backdoor attack for PFL, which removes portions of the zero-frequency (DC) component and replaces them with Gaussian-distributed samples in the frequency domain. Our attack achieves superior attack success rates while maintaining clean accuracy across four datasets (CIFAR-10/100, GTSRB, SVHN) compared to existing spatial-domain attacks, evaluated under parameter decoupling based personalization. DCInject achieves superior performance with ASRs of 96.83% (CIFAR-10), 99.38% (SVHN), and 100% (GTSRB) while maintaining clean accuracy. Under I-BAU defense, DCInject demonstrates strong persistence, retaining 90.30% ASR vs BadNet's 58.56% on VGG-16, exposing critical vulnerabilities in PFL security assumptions. Our code is available at https://github.com/NahomMA/DCINject-PFL --- ### The Million-Label NER: Breaking Scale Barriers with GLiNER bi-encoder Authors: Ihor Stepanov, Mykhailo Shtopko, Dmytro Vodianytskyi, Oleksandr Lukashov Published: 2026-02-11 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2602.18487v1 This paper introduces GLiNER-bi-Encoder, a novel architecture for Named Entity Recognition (NER) that harmonizes zero-shot flexibility with industrial-scale efficiency. While the original GLiNER framework offers strong generalization, its joint-encoding approach suffers from quadratic complexity as the number of entity labels increases. Our proposed bi-encoder design decouples the process into a dedicated label encoder and a context encoder, effectively removing the context-window bottleneck. This architecture enables the simultaneous recognition of thousands, and potentially millions, of entity types with minimal overhead. Experimental results demonstrate state-of-the-art zero-shot performance, achieving 61.5 percent Micro-F1 on the CrossNER benchmark. Crucially, by leveraging pre-computed label embeddings, GLiNER-bi-Encoder achieves up to a 130 times throughput improvement at 1024 labels compared to its uni-encoder predecessors. Furthermore, we introduce GLiNKER, a modular framework that leverages this architecture for high-performance entity linking across massive knowledge bases such as Wikidata. --- ### Support Vector Data Description for Radar Target Detection Authors: Jean Pinsolle, Yadang Alexis Rouzoumka, Chengfang Ren, Chistèle Morisseau, Jean-Philippe Ovarlez Published: 2026-02-11 Categories: cs.LG, eess.SP, stat.ML arXiv: https://arxiv.org/abs/2602.18486v1 Classical radar detection techniques rely on adaptive detectors that estimate the noise covariance matrix from target-free secondary data. While effective in Gaussian environments, these methods degrade in the presence of clutter, which is better modeled by heavy-tailed distributions such as the Complex Elliptically Symmetric (CES) and Compound-Gaussian (CGD) families. Robust covariance estimators like M-estimators or Tyler's estimator address this issue, but still struggle when thermal noise combines with clutter. To overcome these challenges, we investigate the use of Support Vector Data Description (SVDD) and its deep extension, Deep SVDD, for target detection. These one-class learning methods avoid direct noise covariance estimation and are adapted here as CFAR detectors. We propose two novel SVDD-based detection algorithms and demonstrate their effectiveness on simulated radar data. --- ### APEX-SQL: Talking to the data via Agentic Exploration for Text-to-SQL Authors: Bowen Cao, Weibin Liao, Yushi Sun, Dong Fang, Haitao Li, Wai Lam Published: 2026-02-11 Categories: cs.DB, cs.AI arXiv: https://arxiv.org/abs/2602.16720v2 Text-to-SQL systems powered by Large Language Models have excelled on academic benchmarks but struggle in complex enterprise environments. The primary limitation lies in their reliance on static schema representations, which fails to resolve semantic ambiguity and scale effectively to large, complex databases. To address this, we propose APEX-SQL, an Agentic Text-to-SQL Framework that shifts the paradigm from passive translation to agentic exploration. Our framework employs a hypothesis-verification loop to ground model reasoning in real data. In the schema linking phase, we use logical planning to verbalize hypotheses, dual-pathway pruning to reduce the search space, and parallel data profiling to validate column roles against real data, followed by global synthesis to ensure topological connectivity. For SQL generation, we introduce a deterministic mechanism to retrieve exploration directives, allowing the agent to effectively explore data distributions, refine hypotheses, and generate semantically accurate SQLs. Experiments on BIRD (70.65% execution accuracy) and Spider 2.0-Snow (51.01% execution accuracy) demonstrate that APEX-SQL outperforms competitive baselines with reduced token consumption. Further analysis reveals that agentic exploration acts as a performance multiplier, unlocking the latent reasoning potential of foundation models in enterprise settings. Ablation studies confirm the critical contributions of each component in ensuring robust and accurate data analysis. Our code is released at https://github.com/Tencent/APEX-SQL-Project. --- ### MultiCube-RAG for Multi-hop Question Answering Authors: Jimeng Shi, Wei Hu, Runchu Tian, Bowen Jin, Wonbin Kweon, SeongKu Kang, Yunfan Kang, Dingqi Ye, Sizhe Zhou, Shaowen Wang, Jiawei Han Published: 2026-02-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2602.15898v1 Multi-hop question answering (QA) necessitates multi-step reasoning and retrieval across interconnected subjects, attributes, and relations. Existing retrieval-augmented generation (RAG) methods struggle to capture these structural semantics accurately, resulting in suboptimal performance. Graph-based RAGs structure such information in graphs, but the resulting graphs are often noisy and computationally expensive. Moreover, most methods rely on single-step retrieval, neglecting the need for multi-hop reasoning processes. Recent training-based approaches attempt to incentivize the large language models (LLMs) for iterative reasoning and retrieval, but their training processes are prone to unstable convergence and high computational overhead. To address these limitations, we devise an ontology-based cube structure with multiple and orthogonal dimensions to model structural subjects, attributes, and relations. Built on the cube structure, we propose MultiCube-RAG, a training-free method consisting of multiple cubes for multi-step reasoning and retrieval. Each cube specializes in modeling a class of subjects, so that MultiCube-RAG flexibly selects the most suitable cubes to acquire the relevant knowledge precisely. To enhance the query-based reasoning and retrieval, our method decomposes a complex multi-hop query into a set of simple subqueries along cube dimensions and conquers each of them sequentially. Experiments on four multi-hop QA datasets show that MultiCube-RAG improves response accuracy by 8.9% over the average performance of various baselines. Notably, we also demonstrate that our method performs with greater efficiency and inherent explainability. --- ### Mitigating Gradient Inversion Risks in Language Models via Token Obfuscation Authors: Xinguo Feng, Zhongkui Ma, Zihan Wang, Alsharif Abuadbba, Guangdong Bai Published: 2026-02-11 Categories: cs.CL arXiv: https://arxiv.org/abs/2602.15897v1 Training and fine-tuning large-scale language models largely benefit from collaborative learning, but the approach has been proven vulnerable to gradient inversion attacks (GIAs), which allow adversaries to reconstruct private training data from shared gradients. Existing defenses mainly employ gradient perturbation techniques, e.g., noise injection or gradient pruning, to disrupt GIAs' direct mapping from gradient space to token space. However, these methods often fall short due to the retention of semantics similarity across gradient, embedding, and token spaces. In this work, we propose a novel defense mechanism named GHOST (gradient shield with obfuscated tokens), a token-level obfuscation mechanism that neutralizes GIAs by decoupling the inherent connections across gradient, embedding, and token spaces. GHOST is built upon an important insight: due to the large scale of the token space, there exist semantically distinct yet embedding-proximate tokens that can serve as the shadow substitutes of the original tokens, which enables a semantic disconnection in the token space while preserving the connection in the embedding and gradient spaces. GHOST comprises a searching step, which identifies semantically distinct candidate tokens using a multi-criteria searching process, and a selection step, which selects optimal shadow tokens to ensure minimal disruption to features critical for training by preserving alignment with the internal outputs produced by original tokens. Evaluation across diverse model architectures (from BERT to Llama) and datasets demonstrates the remarkable effectiveness of GHOST in protecting privacy (as low as 1% in recovery rate) and preserving utility (up to 0.92 in classification F1 and 5.45 in perplexity), in both classification and generation tasks against state-of-the-art GIAs and adaptive attack scenarios. ---