# Research Papers — 2026-01 · bundle page 98 of 189 > Full-text markdown of papers #4851–4900 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/98 [← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/97.md) [Next bundle →](https://aipod.dev/papers/archive/2026-01/page/99.md) ### AVIR: Adaptive Visual In-Document Retrieval for Efficient Multi-Page Document Question Answering Authors: Zongmin Li, Yachuan Li, Lei Kang, Dimosthenis Karatzas, Wenkang Ma Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11976v1 Multi-page Document Visual Question Answering (MP-DocVQA) remains challenging because long documents not only strain computational resources but also reduce the effectiveness of the attention mechanism in large vision-language models (LVLMs). We tackle these issues with an Adaptive Visual In-document Retrieval (AVIR) framework. A lightweight retrieval model first scores each page for question relevance. Pages are then clustered according to the score distribution to adaptively select relevant content. The clustered pages are screened again by Top-K to keep the context compact. However, for short documents, clustering reliability decreases, so we use a relevance probability threshold to select pages. The selected pages alone are fed to a frozen LVLM for answer generation, eliminating the need for model fine-tuning. The proposed AVIR framework reduces the average page count required for question answering by 70%, while achieving an ANLS of 84.58% on the MP-DocVQA dataset-surpassing previous methods with significantly lower computational cost. The effectiveness of the proposed AVIR is also verified on the SlideVQA and DUDE benchmarks. The code is available at https://github.com/Li-yachuan/AVIR. --- ### Learn Like Humans: Use Meta-cognitive Reflection for Efficient Self-Improvement Authors: Xinmeng Hou, Peiliang Gong, Bohao Qu, Wuqi Wang, Qing Guo, Yang Liu Published: 2026-01-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.11974v1 While Large Language Models (LLMs) enable complex autonomous behavior, current agents remain constrained by static, human-designed prompts that limit adaptability. Existing self-improving frameworks attempt to bridge this gap but typically rely on inefficient, multi-turn recursive loops that incur high computational costs. To address this, we propose Metacognitive Agent Reflective Self-improvement (MARS), a framework that achieves efficient self-evolution within a single recurrence cycle. Inspired by educational psychology, MARS mimics human learning by integrating principle-based reflection (abstracting normative rules to avoid errors) and procedural reflection (deriving step-by-step strategies for success). By synthesizing these insights into optimized instructions, MARS allows agents to systematically refine their reasoning logic without continuous online feedback. Extensive experiments on six benchmarks demonstrate that MARS outperforms state-of-the-art self-evolving systems while significantly reducing computational overhead. --- ### Real-Time Multi-Modal Embedded Vision Framework for Object Detection Facial Emotion Recognition and Biometric Identification on Low-Power Edge Platforms Authors: S. M. Khalid Bin Zahid, Md. Rakibul Hasan Nishat, Abdul Hasib, Md. Rakibul Hasan, Md. Ashiqussalehin, Md. Sahadat Hossen Sajib, A. S. M. Ahsanul Sarkar Akib Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11970v1 Intelligent surveillance systems often handle perceptual tasks such as object detection, facial recognition, and emotion analysis independently, but they lack a unified, adaptive runtime scheduler that dynamically allocates computational resources based on contextual triggers. This limits their holistic understanding and efficiency on low-power edge devices. To address this, we present a real-time multi-modal vision framework that integrates object detection, owner-specific face recognition, and emotion detection into a unified pipeline deployed on a Raspberry Pi 5 edge platform. The core of our system is an adaptive scheduling mechanism that reduces computational load by 65\% compared to continuous processing by selectively activating modules such as, YOLOv8n for object detection, a custom FaceNet-based embedding system for facial recognition, and DeepFace's CNN for emotion classification. Experimental results demonstrate the system's efficacy, with the object detection module achieving an Average Precision (AP) of 0.861, facial recognition attaining 88\% accuracy, and emotion detection showing strong discriminatory power (AUC up to 0.97 for specific emotions), while operating at 5.6 frames per second. Our work demonstrates that context-aware scheduling is the key to unlocking complex multi-modal AI on cost-effective edge hardware, making intelligent perception more accessible and privacy-preserving. --- ### MemoryRewardBench: Benchmarking Reward Models for Long-Term Memory Management in Large Language Models Authors: Zecheng Tang, Baibei Ji, Ruoxi Sun, Haitian Wang, WangJie You, Zhang Yijun, Wenpeng Zhu, Ji Qi, Juntao Li, Min Zhang Published: 2026-01-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.11969v2 Existing works increasingly adopt memory-centric mechanisms to process long contexts in a segment manner, and effective memory management is one of the key capabilities that enables large language models to effectively propagate information across the entire sequence. Therefore, leveraging reward models (RMs) to automatically and reliably evaluate memory quality is critical. In this work, we introduce MemoryRewardBench, the first benchmark to systematically study the ability of RMs to evaluate long-term memory management processes. MemoryRewardBench covers both long-context comprehension and long-form generation tasks, featuring 10 distinct settings with different memory management patterns, with context length ranging from 8K to 128K tokens. Evaluations on 13 cutting-edge RMs indicate a diminishing performance gap between open-source and proprietary models, with newer-generation models consistently outperforming their predecessors regardless of parameter count. We further expose the capabilities and fundamental limitations of current RMs in evaluating LLM memory management across diverse settings. --- ### A Constraint Programming Model for the Super-Agile Earth Observation Satellite Imaging Scheduling Problem Authors: Margarida Caleiras, Samuel Moniz, Paulo Jorge Nascimento Published: 2026-01-17 Categories: eess.SY, cs.CV, math.OC arXiv: https://arxiv.org/abs/2601.11967v2 As the dependence on satellite imaging continues to grow, modern satellites have become increasingly agile, with the new generation, namely super-agile Earth observation satellites (SAEOS), providing unprecedented imaging flexibility. The highly dynamic capabilities of these satellites introduce additional challenges to the scheduling of observation tasks, as existing approaches for conventional agile satellites do not account for variable observation durations and multiple imaging directions. Although some efforts have been made in this regard, the SAEOS imaging scheduling problem (SAEOS-ISP) remains largely unexplored, and no exact approaches have yet been proposed. In this context, this study presents the first exact Constraint Programming formulation for the SAEOS-ISP, considering flexible observation windows, multiple pointing directions and sequence-dependent transition times across multiple satellites. Computational experiments on a newly generated benchmark set demonstrate that the model can be solved efficiently and within very short computational times. Moreover, the results also show that the proposed approach has the potential to achieve higher computational performance compared to the non-exact approaches that are currently considered state-of-the-art. --- ### R$^2$PO: Decoupling Rollout and Inference Policies for LLM Reasoning Authors: Jingchu Wang, Bingbing Xu, Yige Yuan, Dan Zhang, Bin Xie, Xiaoqian Sun, Huawei Shen Published: 2026-01-17 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2601.11960v3 Existing reinforcement learning methods for LLM reasoning implicitly assume that the policy generating training trajectories should coincide with the one producing inference responses. We argue that this is a misleading inductive bias: the optimization-optimal trajectory distribution favors informative gradients, whereas the inference-optimal response distribution emphasizes accuracy and consistency. Forcing both into a single policy entangles their gradients and suppresses exploration. We propose R$^2$PO (Residual Rollout Policy Optimization), which attaches a lightweight Residual Rollout-Head atop the policy to decouple training trajectories from inference responses, diversifying rollouts during training while keeping inference generation intact. Experiments show that R$^2$PO consistently outperforms baselines, with average accuracy gains of 3.4% on MATH-500 and 1.3% on APPS, alongside more diverse rollouts and reduced length bias. Our code is available at https://github.com/RRPO-ARR/Code. --- ### PEARL: Self-Evolving Assistant for Time Management with Reinforcement Learning Authors: Bingxuan Li, Jeonghwan Kim, Cheng Qian, Xiusi Chen, Eitan Anzenberg, Niran Kundapur, Heng Ji Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11957v4 Overlapping calendar invitations force busy professionals to repeatedly decide which meetings to attend, reschedule, or decline. We refer to this preference-driven decision process as calendar conflict resolution. Automating this decision process is crucial yet challenging. Scheduling logistics can drain hours, and human delegation often fails at scale, which motivates us to ask: Can we trust large language models (LLMs) or language agents to manage time? To enable a systematic study of this question, we introduce CalConflictBench, a benchmark for long-horizon calendar conflict resolution. In CalConflictBench, conflicts are presented to agents round-by-round over a calendar year, requiring them to infer and adapt to user preferences progressively. Our experiments show that current LLM agents perform poorly with high error rates, e.g., Qwen-3-30B-Think has an average error rate of 35%. To address this gap, we propose PEARL, a reinforcement-learning framework that (i) augments the language agent with an external preference memory that stores and updates inferred strategies (e.g., attendee priorities, topic importance, time/location preferences), and (ii) optimizes the agent with round-wise rewards that directly supervise decision correctness, ranking quality, and memory usage across rounds. Experiments on CalConflictBench show that PEARL achieves an error reduction rate of 0.76 and a 55% improvement in average error rate compared to the strongest baseline. --- ### Double-Calibration: Towards Reliable LLMs via Calibrating Knowledge and Reasoning Confidence Authors: Yuyin Lu, Ziran Liang, Yanghui Rao, Wenqi Fan, Fu Lee Wang, Qing Li Published: 2026-01-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.11956v2 Reliable reasoning in Large Language Models (LLMs) is challenged by their propensity for hallucination. While augmenting LLMs with Knowledge Graphs (KGs) improves factual accuracy, existing KG-augmented methods fail to quantify epistemic uncertainty in both the retrieved evidence and LLMs' reasoning. To bridge this gap, we introduce DoublyCal, a framework built on a novel double-calibration principle. DoublyCal employs a lightweight proxy model to first generate KG evidence alongside a calibrated evidence confidence. This calibrated supporting evidence then guides a black-box LLM, yielding final predictions that are not only more accurate but also well-calibrated, with confidence scores traceable to the uncertainty of the supporting evidence. Experiments on knowledge-intensive benchmarks show that DoublyCal significantly improves both the accuracy and confidence calibration of black-box LLMs while maintaining low token cost. --- ### Data-centric Prompt Tuning for Dynamic Graphs Authors: Yufei Peng, Cheng Yang, Zhengjie Fan, Chuan Shi Published: 2026-01-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.11954v1 Dynamic graphs have attracted increasing attention due to their ability to model complex and evolving relationships in real-world scenarios. Traditional approaches typically pre-train models using dynamic link prediction and directly apply the resulting node temporal embeddings to specific downstream tasks. However, the significant differences among downstream tasks often lead to performance degradation, especially under few-shot settings. Prompt tuning has emerged as an effective solution to this problem. Existing prompting methods are often strongly coupled with specific model architectures or pretraining tasks, which makes it difficult to adapt to recent or future model designs. Moreover, their exclusive focus on modifying node or temporal features while neglecting spatial structural information leads to limited expressiveness and degraded performance. To address these limitations, we propose DDGPrompt, a data-centric prompting framework designed to effectively refine pre-trained node embeddings at the input data level, enabling better adaptability to diverse downstream tasks. We first define a unified node expression feature matrix that aggregates all relevant temporal and structural information of each node, ensuring compatibility with a wide range of dynamic graph models. Then, we introduce three prompt matrices (temporal bias, edge weight, and feature mask) to adjust the feature matrix completely, achieving task-specific adaptation of node embeddings. We evaluate DDGPrompt under a strict few-shot setting on four public dynamic graph datasets. Experimental results demonstrate that our method significantly outperforms traditional methods and prompting approaches in scenarios with limited labels and cold-start conditions. --- ### Controlling Underestimation Bias in Constrained Reinforcement Learning for Safe Exploration Authors: Shiqing Gao, Jiaxin Ding, Luoyi Fu, Xinbing Wang Published: 2026-01-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.11953v1 Constrained Reinforcement Learning (CRL) aims to maximize cumulative rewards while satisfying constraints. However, existing CRL algorithms often encounter significant constraint violations during training, limiting their applicability in safety-critical scenarios. In this paper, we identify the underestimation of the cost value function as a key factor contributing to these violations. To address this issue, we propose the Memory-driven Intrinsic Cost Estimation (MICE) method, which introduces intrinsic costs to mitigate underestimation and control bias to promote safer exploration. Inspired by flashbulb memory, where humans vividly recall dangerous experiences to avoid risks, MICE constructs a memory module that stores previously explored unsafe states to identify high-cost regions. The intrinsic cost is formulated as the pseudo-count of the current state visiting these risk regions. Furthermore, we propose an extrinsic-intrinsic cost value function that incorporates intrinsic costs and adopts a bias correction strategy. Using this function, we formulate an optimization objective within the trust region, along with corresponding optimization methods. Theoretically, we provide convergence guarantees for the proposed cost value function and establish the worst-case constraint violation for the MICE update. Extensive experiments demonstrate that MICE significantly reduces constraint violations while preserving policy performance comparable to baselines. --- ### Decoder Gradient Shields: A Family of Provable and High-Fidelity Methods Against Gradient-Based Box-Free Watermark Removal Authors: Haonan An, Guang Hua, Wei Du, Hangcheng Cao, Yihang Tao, Guowen Xu, Susanto Rahardja, Yuguang Fang Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11952v1 Box-free model watermarking has gained significant attention in deep neural network (DNN) intellectual property protection due to its model-agnostic nature and its ability to flexibly manage high-entropy image outputs from generative models. Typically operating in a black-box manner, it employs an encoder-decoder framework for watermark embedding and extraction. While existing research has focused primarily on the encoders for the robustness to resist various attacks, the decoders have been largely overlooked, leading to attacks against the watermark. In this paper, we identify one such attack against the decoder, where query responses are utilized to obtain backpropagated gradients to train a watermark remover. To address this issue, we propose Decoder Gradient Shields (DGSs), a family of defense mechanisms, including DGS at the output (DGS-O), at the input (DGS-I), and in the layers (DGS-L) of the decoder, with a closed-form solution for DGS-O and provable performance for all DGS. Leveraging the joint design of reorienting and rescaling of the gradients from watermark channel gradient leaking queries, the proposed DGSs effectively prevent the watermark remover from achieving training convergence to the desired low-loss value, while preserving image quality of the decoder output. We demonstrate the effectiveness of our proposed DGSs in diverse application scenarios. Our experimental results on deraining and image generation tasks with the state-of-the-art box-free watermarking show that our DGSs achieve a defense success rate of 100% under all settings. --- ### Deep learning-based neurodevelopmental assessment in preterm infants Authors: Lexin Ren, Jiamiao Lu, Weichuan Zhang, Benqing Wu, Tuo Wang, Yi Liao, Jiapan Guo, Changming Sun, Liang Guo Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11944v1 Preterm infants (born between 28 and 37 weeks of gestation) face elevated risks of neurodevelopmental delays, making early identification crucial for timely intervention. While deep learning-based volumetric segmentation of brain MRI scans offers a promising avenue for assessing neonatal neurodevelopment, achieving accurate segmentation of white matter (WM) and gray matter (GM) in preterm infants remains challenging due to their comparable signal intensities (isointense appearance) on MRI during early brain development. To address this, we propose a novel segmentation neural network, named Hierarchical Dense Attention Network. Our architecture incorporates a 3D spatial-channel attention mechanism combined with an attention-guided dense upsampling strategy to enhance feature discrimination in low-contrast volumetric data. Quantitative experiments demonstrate that our method achieves superior segmentation performance compared to state-of-the-art baselines, effectively tackling the challenge of isointense tissue differentiation. Furthermore, application of our algorithm confirms that WM and GM volumes in preterm infants are significantly lower than those in term infants, providing additional imaging evidence of the neurodevelopmental delays associated with preterm birth. The code is available at: https://github.com/ICL-SUST/HDAN. --- ### Trainability-Oriented Hybrid Quantum Regression via Geometric Preconditioning and Curriculum Optimization Authors: Qingyu Meng, Yangshuai Wang Published: 2026-01-17 Categories: cs.LG, quant-ph arXiv: https://arxiv.org/abs/2601.11942v4 Quantum neural networks (QNNs) have attracted growing interest for scientific machine learning, yet in regression settings they often suffer from limited trainability under noisy gradients and ill-conditioned optimization. We propose a hybrid quantum--classical regression framework designed to mitigate these bottlenecks. Our model prepends a lightweight classical embedding that acts as a learnable geometric preconditioner, reshaping the input representation to better condition a downstream variational quantum circuit. Building on this architecture, we introduce a curriculum optimization protocol that progressively increases circuit depth and transitions from SPSA-based stochastic exploration to Adam-based gradient fine-tuning. We evaluate the approach on PDE-informed regression benchmarks and standard regression datasets under a fixed training budget in a simulator setting. Empirically, the proposed framework consistently improves over pure QNN baselines and yields more stable convergence in data-limited regimes. We further observe reduced structured errors that are visually correlated with oscillatory components on several scientific benchmarks, suggesting that geometric preconditioning combined with curriculum training is a practical approach for stabilizing quantum regression. --- ### Thinking Traps in Long Chain-of-Thought: A Measurable Study and Trap-Aware Adaptive Restart Authors: Kang Chen, Fan Yu, Junjie Nian, Shihan Zhao, Zhuoka Feng, Zijun Yao, Heng Wang, Minshen Yu, Yixin Cao Published: 2026-01-17 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2601.11940v1 Scaling test-time compute via Long Chain-of-Thought (Long-CoT) significantly enhances reasoning capabilities, yet extended generation does not guarantee correctness: after an early wrong commitment, models may keep elaborating a self-consistent but incorrect prefix. Through fine-grained trajectory analysis, we identify Thinking Traps, prefix-dominant deadlocks where later reflection, alternative attempts, or verification fails to revise the root error. On a curated subset of DAPO-MATH, 89\% of failures exhibit such traps. To solve this problem, we introduce TAAR (Trap-Aware Adaptive Restart), a test-time control framework that trains a diagnostic policy to predict two signals from partial trajectories: a trap index for where to truncate and an escape probability for whether and how strongly to intervene. At inference time, TAAR truncates the trajectory before the predicted trap segment and adaptively restarts decoding; for severely trapped cases, it applies stronger perturbations, including higher-temperature resampling and an optional structured reboot suffix. Experiments on challenging mathematical and scientific reasoning benchmarks (AIME24, AIME25, GPQA-Diamond, HMMT25, BRUMO25) show that TAAR improves reasoning performance without fine-tuning base model parameters. --- ### Impact of Circuit Depth versus Qubit Count on Variational Quantum Classifiers for Higgs Boson Signal Detection Authors: Fatih Maulana Published: 2026-01-17 Categories: quant-ph, cs.LG, hep-ex arXiv: https://arxiv.org/abs/2601.11937v1 High-Energy Physics (HEP) experiments, such as those at the Large Hadron Collider (LHC), generate massive datasets that challenge classical computational limits. Quantum Machine Learning (QML) offers a potential advantage in processing high-dimensional data; however, finding the optimal architecture for current Noisy Intermediate-Scale Quantum (NISQ) devices remains an open challenge. This study investigates the performance of Variational Quantum Classifiers (VQC) in detecting Higgs Boson signals using the ATLAS Higgs Boson Machine Learning Challenge 2014 experiment dataset. We implemented a dimensionality reduction pipeline using Principal Component Analysis (PCA) to map 30 physical features into 4-qubit and 8-qubit latent spaces. We benchmarked three configurations: (A) a shallow 4-qubit circuit, (B) a deep 4-qubit circuit with increased entanglement layers, and (C) an expanded 8-qubit circuit. Experimental results demonstrate that increasing circuit depth significantly improves performance, yielding the highest accuracy of 56.2% (Configuration B), compared to a baseline of 51.9%. Conversely, simply scaling to 8 qubits resulted in a performance degradation to 50.6% due to optimization challenges associated with Barren Plateaus in the larger Hilbert space. These findings suggest that for near-term quantum hardware, prioritizing circuit depth and entanglement capability is more critical than increasing qubit count for effective anomaly detection in HEP data. --- ### Big Data Workload Profiling for Energy-Aware Cloud Resource Management Authors: Milan Parikh, Aniket Abhishek Soni, Sneja Mitinbhai Shah, Ayush Raj Jha Published: 2026-01-17 Categories: cs.DC, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2601.11935v1 Cloud data centers face increasing pressure to reduce operational energy consumption as big data workloads continue to grow in scale and complexity. This paper presents a workload aware and energy efficient scheduling framework that profiles CPU utilization, memory demand, and storage IO behavior to guide virtual machine placement decisions. By combining historical execution logs with real time telemetry, the proposed system predicts the energy and performance impact of candidate placements and enables adaptive consolidation while preserving service level agreement compliance. The framework is evaluated using representative Hadoop MapReduce, Spark MLlib, and ETL workloads deployed on a multi node cloud testbed. Experimental results demonstrate consistent energy savings of 15 to 20 percent compared to a baseline scheduler, with negligible performance degradation. These findings highlight workload profiling as a practical and scalable strategy for improving the sustainability of cloud based big data processing environments. --- ### Event Detection with a Context-Aware Encoder and LoRA for Improved Performance on Long-Tailed Classes Authors: Abdullah Al Monsur, Nitesh Vamshi Bommisetty, Gene Louis Kim Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11932v2 The current state of event detection research has two notable re-occurring limitations that we investigate in this study. First, the unidirectional nature of decoder-only LLMs presents a fundamental architectural bottleneck for natural language understanding tasks that depend on rich, bidirectional context. Second, we confront the conventional reliance on Micro-F1 scores in event detection literature, which systematically inflates performance by favoring majority classes. Instead, we focus on Macro-F1 as a more representative measure of a model's ability across the long-tail of event types. Our experiments demonstrate that models enhanced with sentence context achieve superior performance over canonical decoder-only baselines. Using Low-Rank Adaptation (LoRA) during finetuning provides a substantial boost in Macro-F1 scores in particular, especially for the decoder-only models, showing that LoRA can be an effective tool to enhance LLMs' performance on long-tailed event classes. --- ### Language-Guided and Motion-Aware Gait Representation for Generalizable Recognition Authors: Zhengxian Wu, Chuanrui Zhang, Shenao Jiang, Hangrui Xu, Zirui Liao, Luyuan Zhang, Huaqiu Li, Peng Jiao, Haoqian Wang Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11931v2 Gait recognition is emerging as a promising technology and an innovative field within computer vision, with a wide range of applications in remote human identification. However, existing methods typically rely on complex architectures to directly extract features from images and apply pooling operations to obtain sequence-level representations. Such designs often lead to overfitting on static noise (e.g., clothing), while failing to effectively capture dynamic motion regions, such as the arms and legs. This bottleneck is particularly challenging in the presence of intra-class variation, where gait features of the same individual under different environmental conditions are significantly distant in the feature space. To address the above challenges, we present a Languageguided and Motion-aware gait recognition framework, named LMGait. To the best of our knowledge, LMGait is the first method to introduce natural language descriptions as explicit semantic priors into the gait recognition task. In particular, we utilize designed gait-related language cues to capture key motion features in gait sequences. To improve cross-modal alignment, we propose the Motion Awareness Module (MAM), which refines the language features by adaptively adjusting various levels of semantic information to ensure better alignment with the visual representations. Furthermore, we introduce the Motion Temporal Capture Module (MTCM) to enhance the discriminative capability of gait features and improve the model's motion tracking ability. We conducted extensive experiments across multiple datasets, and the results demonstrate the significant advantages of our proposed network. Specifically, our model achieved accuracies of 88.5%, 97.1%, and 97.5% on the CCPG, SUSTech1K, and CASIAB datasets, respectively, achieving state-of-the-art performance. Homepage: https://dingwu1021.github.io/LMGait/ --- ### SupScene: Scene-Structured Overlap Supervision for Image Retrieval in Unconstrained SfM Authors: Xulei Shi, Maoyu Wang, Yuning Peng, Guanbo Wang, Xin Wang, Yifan Liao, Qi Chen, Pengjie Tao Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11930v2 Image retrieval is a critical step for reducing the quadratic cost of image matching in unconstrained Structure-from-Motion (SfM). Unlike generic image retrieval, however, the relevant goal of SfM is to identify geometrically matchable image pairs rather than merely semantically similar images. Prevailing methods are largely trained under anchor-centric tuple guidance, which organizes the training around isolated tuples and under-utilizes the dense, graded overlap structure naturally established within a SfM scene. In this work, we present SupScene, a scene-structured training framework that samples connected local subgraphs from SfM overlap graphs and jointly supervises all valid within-subgraph pairwise relations. To explicitly align the trained descriptor with geometric co-visibility, we further introduce an overlap-ordered objective that combines multi-similarity optimization with a continuous relative-overlap ranking term. In addition, the proposed framework is instantiated with a lightweight Structural Context Probe Pooling (SCPP) head that aggregates complementary structural responses into a compact global descriptor. Extensive experimental results on multiple benchmarks demonstrate that our method can significantly improve overall retrieval performance and enhance the completeness of downstream SfM reconstructions. Code and models are available at https://github.com/Suxilan/SupScene. --- ### Harmonica: A Self-Adaptation Exemplar for Sustainable MLOps Authors: Ananya Halgatti, Shaunak Biswas, Hiya Bhatt, Srinivasan Rakhunathan, Karthik Vaidhyanathan Published: 2026-01-17 Categories: cs.SE, cs.LG arXiv: https://arxiv.org/abs/2601.11926v2 Machine learning enabled systems (MLS) often operate in settings where they regularly encounter uncertainties arising from changes in their surrounding environment. Without structured oversight, such changes can degrade model behavior, increase operational cost, and reduce the usefulness of deployed systems. Although Machine Learning Operations (MLOps) streamlines the lifecycle of ML models, it provides limited support for addressing runtime uncertainties that influence the longer term sustainability of MLS. To support continued viability, these systems need a mechanism that detects when execution drifts outside acceptable bounds and adjusts system behavior in response. Despite the growing interest in sustainable and self-adaptive MLS, there has been limited work towards exemplars that allow researchers to study these challenges in MLOps pipelines. This paper presents Harmonica, a self-adaptation exemplar built on the HarmonE approach, designed to enable the sustainable operation of such pipelines. Harmonica introduces structured adaptive control through MAPE-K loop, separating high-level adaptation policy from low-level tactic execution. It continuously monitors sustainability metrics, evaluates them against dynamic adaptation boundaries, and automatically triggers architectural tactics when thresholds are violated. We demonstrate the tool through case studies in time series regression and computer vision, examining its ability to improve system stability and reduce manual intervention. The results show that Harmonica offers a practical and reusable foundation for enabling adaptive behavior in MLS that rely on MLOps pipelines for sustained operation. --- ### Communication-Corruption Coupling and Verification in Cooperative Multi-Objective Bandits Authors: Ming Shi Published: 2026-01-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.11924v2 We study cooperative stochastic multi-armed bandits with vector-valued rewards under adversarial corruption and limited verification. In each of $T$ rounds, each of $N$ agents selects an arm, the environment generates a clean reward vector, and an adversary perturbs the observed feedback subject to a global corruption budget $Γ$. Performance is measured by team regret under a coordinate-wise nondecreasing, $L$-Lipschitz scalarization $φ$, covering linear, Chebyshev, and smooth monotone utilities. Our main contribution is a communication-corruption coupling: we show that a fixed environment-side budget $Γ$ can translate into an effective corruption level ranging from $Γ$ to $NΓ$, depending on whether agents share raw samples, sufficient statistics, or only arm recommendations. We formalize this via a protocol-induced multiplicity functional and prove regret bounds parameterized by the resulting effective corruption. As corollaries, raw-sample sharing can suffer an $N$-fold larger additive corruption penalty, whereas summary sharing and recommendation-only sharing preserve an unamplified $O(Γ)$ term and achieve centralized-rate team regret. We further establish information-theoretic limits, including an unavoidable additive $Ω(Γ)$ penalty and a high-corruption regime $Γ=Θ(NT)$ where sublinear regret is impossible without clean information. Finally, we characterize how a global budget $ν$ of verified observations restores learnability. That is, verification is necessary in the high-corruption regime, and sufficient once it crosses the identification threshold, with certified sharing enabling the team's regret to become independent of $Γ$. --- ### Mapping the maturation of TCM as an adjuvant to radiotherapy Authors: P. Bilha Githinji, Aikaterini Melliou, Xi Yuan, Dayan Zhang, Lian Zhang, Zhenglin Chen, Jiansong Ji, Chengying Lv, Jinhao Xu, Peiwu Qin, Dongmei Yu Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11923v3 The integration of complementary medicine into oncology represents a paradigm shift that has seen to increasing adoption of Traditional Chinese Medicine (TCM) as an adjuvant to radiotherapy. About twenty-five years since the formal institutionalization of integrated oncology, it is opportune to synthesize the trajectory of evidence for TCM as an adjuvant to radiotherapy. Here we conduct a large-scale analysis of 69,745 publications (2000 - 2025), emerging a cyclical evolution defined by coordinated expansion and contraction in publication output, international collaboration, and funding commitments that mirrors a define-ideate-test pattern. Using a theme modeling workflow designed to determine a stable thematic structure of the field, we identify five dominant thematic axes - cancer types, supportive care, clinical endpoints, mechanisms, and methodology - that signal a focus on patient well-being, scientific rigor and mechanistic exploration. Cross-theme integration of TCM is patient-centered and systems-oriented. Together with the emergent cycles of evolution, the thematic structure demonstrates progressive specialization and potential defragmentation of the field or saturation of existing research agenda. The analysis points to a field that has matured its current research agenda and is likely at the cusp of something new. Additionally, the field exhibits positive reporting of findings that is homogeneous across publication types, thematic areas, and the cycles of evolution suggesting a system-wide positive reporting bias agnostic to structural drivers. --- ### Enhancing LLM-Based Data Annotation with Error Decomposition Authors: Zhen Xu, Vedant Khatri, Yijun Dai, Xiner Liu, Siyan Li, Xuanming Zhang, Renzhe Yu Published: 2026-01-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.11920v1 Large language models offer a scalable alternative to human coding for data annotation tasks, enabling the scale-up of research across data-intensive domains. While LLMs are already achieving near-human accuracy on objective annotation tasks, their performance on subjective annotation tasks, such as those involving psychological constructs, is less consistent and more prone to errors. Standard evaluation practices typically collapse all annotation errors into a single alignment metric, but this simplified approach may obscure different kinds of errors that affect final analytical conclusions in different ways. Here, we propose a diagnostic evaluation paradigm that incorporates a human-in-the-loop step to separate task-inherent ambiguity from model-driven inaccuracies and assess annotation quality in terms of their potential downstream impacts. We refine this paradigm on ordinal annotation tasks, which are common in subjective annotation. The refined paradigm includes: (1) a diagnostic taxonomy that categorizes LLM annotation errors along two dimensions: source (model-specific vs. task-inherent) and type (boundary ambiguity vs. conceptual misidentification); (2) a lightweight human annotation test to estimate task-inherent ambiguity from LLM annotations; and (3) a computational method to decompose observed LLM annotation errors following our taxonomy. We validate this paradigm on four educational annotation tasks, demonstrating both its conceptual validity and practical utility. Theoretically, our work provides empirical evidence for why excessively high alignment is unrealistic in specific annotation tasks and why single alignment metrics inadequately reflect the quality of LLM annotations. In practice, our paradigm can be a low-cost diagnostic tool that assesses the suitability of a given task for LLM annotation and provides actionable insights for further technical optimization. --- ### Effects of Gabor Filters on Classification Performance of CNNs Trained on a Limited Number of Conditions Authors: Akito Morita, Hirotsugu Okuno Published: 2026-01-17 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.11918v1 In this study, we propose a technique to improve the accuracy and reduce the size of convolutional neural networks (CNNs) running on edge devices for real-world robot vision applications. CNNs running on edge devices must have a small architecture, and CNNs for robot vision applications involving on-site object recognition must be able to be trained efficiently to identify specific visual targets from data obtained under a limited variation of conditions. The visual nervous system (VNS) is a good example that meets the above requirements because it learns from few visual experiences. Therefore, we used a Gabor filter, a model of the feature extractor of the VNS, as a preprocessor for CNNs to investigate the accuracy of the CNNs trained with small amounts of data. To evaluate how well CNNs trained on image data acquired under a limited variation of conditions generalize to data acquired under other conditions, we created an image dataset consisting of images acquired from different camera positions, and investigated the accuracy of the CNNs that trained using images acquired at a certain distance. The results were compared after training on multiple CNN architectures with and without Gabor filters as preprocessing. The results showed that preprocessing with Gabor filters improves the generalization performance of CNNs and contributes to reducing the size of CNNs. --- ### Low-rank Orthogonal Subspace Intervention for Generalizable Face Forgery Detection Authors: Chi Wang, Xinjue Hu, Boyu Wang, Ziwen He, Zhangjie Fu Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11915v2 The generalization problem remains a key challenge in face forgery detection. This paper explores the reasons for the generalization failure of Vanilla CLIP: in ``real vs. fake" detection, the few dominant principal components in the feature space primarily encode forgery-irrelevant information, rather than authentic forgery traces. However, this irrelevant information inevitably leads to spurious correlations, severely limiting detector performance. We define this phenomenon as ``low-rank spurious bias". To address this, we propose a low-rank representation space intervention paradigm, named the SeLop, from the perspective of causal representation learning. SeLop unifies the spurious correlation factors irrelevant to forgery into a low-rank subspace and cuts off the statistical shortcut between it and the label, thus aligning representation learning with authentic forgery traces. Specifically, we decompose spurious correlation features into a low-rank subspace through orthogonal low-rank projection, then remove this subspace from the original representation and train its orthogonal complement to capture forgery-related features. This low-rank projection removal effectively eliminates spurious correlation factors, ensuring that classification decisions are based on authentic forgery cues. With only 0.39M trainable parameters, our method achieves state-of-the-art performance across several benchmarks, demonstrating excellent robustness and generalization. --- ### LSTM-MAS: A Long Short-Term Memory Inspired Multi-Agent System for Long-Context Understanding Authors: Yichen Jiang, Jiakang Yuan, Chongjun Tu, Peng Ye, Tao Chen Published: 2026-01-17 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.11913v2 Effectively processing long contexts remains a fundamental yet unsolved challenge for large language models (LLMs). Existing single-LLM-based methods primarily reduce the context window or optimize the attention mechanism, but they often encounter additional computational costs or constrained expanded context length. While multi-agent-based frameworks can mitigate these limitations, they remain susceptible to the accumulation of errors and the propagation of hallucinations. In this work, we draw inspiration from the Long Short-Term Memory (LSTM) architecture to design a Multi-Agent System called LSTM-MAS, emulating LSTM's hierarchical information flow and gated memory mechanisms for long-context understanding. Specifically, LSTM-MAS organizes agents in a chained architecture, where each node comprises a worker agent for segment-level comprehension, a filter agent for redundancy reduction, a judge agent for continuous error detection, and a manager agent for globally regulates information propagation and retention, analogous to LSTM and its input gate, forget gate, constant error carousel unit, and output gate. These novel designs enable controlled information transfer and selective long-term dependency modeling across textual segments, which can effectively avoid error accumulation and hallucination propagation. We conducted an extensive evaluation of our method. Compared with the previous best multi-agent approach, CoA, our model achieves improvements of 97.97%, 65.75%, 122.19%, 39.61% and 10.80% on Narrative QA, Qasper, HotpotQA, 2WikiMQA and MuSiQue, respectively. --- ### Reliable Deep Learning for Small-Scale Classifications: Experiments on Real-World Image Datasets from Bangladesh Authors: Alfe Suny, MD Sakib Ul Islam, Md. Imran Hossain Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11911v2 Convolutional neural networks (CNNs) have achieved state-of-the-art performance in image recognition tasks but often involve complex architectures that may overfit on small datasets. In this study, we evaluate a compact CNN across five publicly available, real-world image datasets from Bangladesh, including urban encroachment, vehicle detection, road damage, and agricultural crops. The network demonstrates high classification accuracy, efficient convergence, and low computational overhead. Quantitative metrics and saliency analyses indicate that the model effectively captures discriminative features and generalizes robustly across diverse scenarios, highlighting the suitability of streamlined CNN architectures for small-class image classification tasks. --- ### A Training-Free Guess What Vision Language Model from Snippets to Open-Vocabulary Object Detection Authors: Guiying Zhu, Bowen Yang, Yin Zhuang, Tong Zhang, Guanqun Wang, Zhihao Che, He Chen, Lianlin Li Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11910v2 Open-Vocabulary Object Detection (OVOD) aims to develop the capability to detect anything. Although myriads of large-scale pre-training efforts have built versatile foundation models that exhibit impressive zero-shot capabilities to facilitate OVOD, the necessity of creating a universal understanding for any object cognition according to already pretrained foundation models is usually overlooked. Therefore, in this paper, a training-free Guess What Vision Language Model, called GW-VLM, is proposed to form a universal understanding paradigm based on our carefully designed Multi-Scale Visual Language Searching (MS-VLS) coupled with Contextual Concept Prompt (CCP) for OVOD. This approach can engage a pre-trained Vision Language Model (VLM) and a Large Language Model (LLM) in the game of "guess what". Wherein, MS-VLS leverages multi-scale visual-language soft-alignment for VLM to generate snippets from the results of class-agnostic object detection, while CCP can form the concept of flow referring to MS-VLS and then make LLM understand snippets for OVOD. Finally, the extensive experiments are carried out on natural and remote sensing datasets, including COCO val, Pascal VOC, DIOR, and NWPU-10, and the results indicate that our proposed GW-VLM can achieve superior OVOD performance compared to the-state-of-the-art methods without any training step. --- ### Effects of the retina-inspired light intensity encoding on color discrimination performance Authors: Io Yamada, Hirotsugu Okuno Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11909v1 Color is an important source of information for visual functions such as object recognition, but it is greatly affected by the color of illumination. The ability to perceive the color of a visual target independent of illumination color is called color constancy (CC), and is an important feature for vision systems that use color information. In this study, we investigated the effects of the light intensity encoding function on the performance of CC of the center/surround (C/S) retinex model, which is a well-known model inspired by CC of the visual nervous system. The functions used to encode light intensity are the logarithmic function used in the original C/S retinex model and the Naka-Rushton (N-R) function, which is a model of retinal photoreceptor response. Color-variable LEDs were used to illuminate visual targets with various lighting colors, and color information computed by each model was used to evaluate the degree to which the color of visual targets illuminated with different lighting colors could be discriminated. Color information was represented using the HSV color space and a color plane based on the classical opponent color theory. The results showed that the combination of the N-R function and the double opponent color plane representation provided superior discrimination performance. --- ### PPA-Plan: Proactive Pitfall Avoidance for Reliable Planning in Long-Context LLM Reasoning Authors: Byeongjin Kim, Gyuwan Kim, Seo Yeon Park Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11908v2 Large language models (LLMs) struggle with reasoning over long contexts where relevant information is sparsely distributed. Although plan-and-execute frameworks mitigate this by decomposing tasks into planning and execution, their effectiveness is often limited by unreliable plan generation due to dependence on surface-level cues. Consequently, plans may be based on incorrect assumptions, and once a plan is formed, identifying what went wrong and revising it reliably becomes difficult, limiting the effectiveness of reactive refinement. To address this limitation, we propose PPA-Plan, a proactive planning strategy for long-context reasoning that focuses on preventing such failures before plan generation. PPA-Plan identifies potential logical pitfalls and false assumptions, formulates them as negative constraints, and conditions plan generation on explicitly avoiding these constraints. Experiments on long-context QA benchmarks show that executing plans generated by PPA-Plan consistently outperforms existing plan-and-execute methods and direct prompting. --- ### Towards Airborne Object Detection: A Deep Learning Analysis Authors: Prosenjit Chatterjee, ANK Zaman Published: 2026-01-17 Categories: cs.CV, cs.AI, cs.LG, cs.SE arXiv: https://arxiv.org/abs/2601.11907v1 The rapid proliferation of airborne platforms, including commercial aircraft, drones, and UAVs, has intensified the need for real-time, automated threat assessment systems. Current approaches depend heavily on manual monitoring, resulting in limited scalability and operational inefficiencies. This work introduces a dual-task model based on EfficientNetB4 capable of performing airborne object classification and threat-level prediction simultaneously. To address the scarcity of clean, balanced training data, we constructed the AODTA Dataset by aggregating and refining multiple public sources. We benchmarked our approach on both the AVD Dataset and the newly developed AODTA Dataset and further compared performance against a ResNet-50 baseline, which consistently underperformed EfficientNetB4. Our EfficientNetB4 model achieved 96% accuracy in object classification and 90% accuracy in threat-level prediction, underscoring its promise for applications in surveillance, defense, and airspace management. Although the title references detection, this study focuses specifically on classification and threat-level inference using pre-localized airborne object images provided by existing datasets. --- ### LIBRA: Language Model Informed Bandit Recourse Algorithm for Personalized Treatment Planning Authors: Junyu Cao, Ruijiang Gao, Esmaeil Keyvanshokooh, Jianhao Ma Published: 2026-01-17 Categories: cs.AI, cs.LG, math.ST arXiv: https://arxiv.org/abs/2601.11905v1 We introduce a unified framework that seamlessly integrates algorithmic recourse, contextual bandits, and large language models (LLMs) to support sequential decision-making in high-stakes settings such as personalized medicine. We first introduce the recourse bandit problem, where a decision-maker must select both a treatment action and a feasible, minimal modification to mutable patient features. To address this problem, we develop the Generalized Linear Recourse Bandit (GLRB) algorithm. Building on this foundation, we propose LIBRA, a Language Model-Informed Bandit Recourse Algorithm that strategically combines domain knowledge from LLMs with the statistical rigor of bandit learning. LIBRA offers three key guarantees: (i) a warm-start guarantee, showing that LIBRA significantly reduces initial regret when LLM recommendations are near-optimal; (ii) an LLM-effort guarantee, proving that the algorithm consults the LLM only $O(\log^2 T)$ times, where $T$ is the time horizon, ensuring long-term autonomy; and (iii) a robustness guarantee, showing that LIBRA never performs worse than a pure bandit algorithm even when the LLM is unreliable. We further establish matching lower bounds that characterize the fundamental difficulty of the recourse bandit problem and demonstrate the near-optimality of our algorithms. Experiments on synthetic environments and a real hypertension-management case study confirm that GLRB and LIBRA improve regret, treatment quality, and sample efficiency compared with standard contextual bandits and LLM-only benchmarks. Our results highlight the promise of recourse-aware, LLM-assisted bandit algorithms for trustworthy LLM-bandits collaboration in personalized high-stakes decision-making. --- ### AEMA: Verifiable Evaluation Framework for Trustworthy and Controlled Agentic LLM Systems Authors: YenTing Lee, Keerthi Koneru, Zahra Moslemi, Sheethal Kumar, Ramesh Radhakrishnan Published: 2026-01-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.11903v1 Evaluating large language model (LLM)-based multi-agent systems remains a critical challenge, as these systems must exhibit reliable coordination, transparent decision-making, and verifiable performance across evolving tasks. Existing evaluation approaches often limit themselves to single-response scoring or narrow benchmarks, which lack stability, extensibility, and automation when deployed in enterprise settings at multi-agent scale. We present AEMA (Adaptive Evaluation Multi-Agent), a process-aware and auditable framework that plans, executes, and aggregates multi-step evaluations across heterogeneous agentic workflows under human oversight. Compared to a single LLM-as-a-Judge, AEMA achieves greater stability, human alignment, and traceable records that support accountable automation. Our results on enterprise-style agent workflows simulated using realistic business scenarios demonstrate that AEMA provides a transparent and reproducible pathway toward responsible evaluation of LLM-based multi-agent systems. Keywords Agentic AI, Multi-Agent Systems, Trustworthy AI, Verifiable Evaluation, Human Oversight --- ### RemoteVAR: Autoregressive Visual Modeling for Remote Sensing Change Detection Authors: Yilmaz Korkmaz, Vishal M. Patel Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11898v1 Remote sensing change detection aims to localize and characterize scene changes between two time points and is central to applications such as environmental monitoring and disaster assessment. Meanwhile, visual autoregressive models (VARs) have recently shown impressive image generation capability, but their adoption for pixel-level discriminative tasks remains limited due to weak controllability, suboptimal dense prediction performance and exposure bias. We introduce RemoteVAR, a new VAR-based change detection framework that addresses these limitations by conditioning autoregressive prediction on multi-resolution fused bi-temporal features via cross-attention, and by employing an autoregressive training strategy designed specifically for change map prediction. Extensive experiments on standard change detection benchmarks show that RemoteVAR delivers consistent and significant improvements over strong diffusion-based and transformer-based baselines, establishing a competitive autoregressive alternative for remote sensing change detection. Code will be available \href{https://github.com/yilmazkorkmaz1/RemoteVAR}{\underline{here}}. --- ### Task-tailored Pre-processing: Fair Downstream Supervised Learning Authors: Jinwon Sohn, Guang Lin, Qifan Song Published: 2026-01-17 Categories: cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2601.11897v1 Fairness-aware machine learning has recently attracted various communities to mitigate discrimination against certain societal groups in data-driven tasks. For fair supervised learning, particularly in pre-processing, there have been two main categories: data fairness and task-tailored fairness. The former directly finds an intermediate distribution among the groups, independent of the type of the downstream model, so a learned downstream classification/regression model returns similar predictive scores to individuals inputting the same covariates irrespective of their sensitive attributes. The latter explicitly takes the supervised learning task into account when constructing the pre-processing map. In this work, we study algorithmic fairness for supervised learning and argue that the data fairness approaches impose overly strong regularization from the perspective of the HGR correlation. This motivates us to devise a novel pre-processing approach tailored to supervised learning. We account for the trade-off between fairness and utility in obtaining the pre-processing map. Then we study the behavior of arbitrary downstream supervised models learned on the transformed data to find sufficient conditions to guarantee their fairness improvement and utility preservation. To our knowledge, no prior work in the branch of task-tailored methods has theoretically investigated downstream guarantees when using pre-processed data. We further evaluate our framework through comparison studies based on tabular and image data sets, showing the superiority of our framework which preserves consistent trade-offs among multiple downstream models compared to recent competing models. Particularly for computer vision data, we see our method alters only necessary semantic features related to the central machine learning task to achieve fairness. --- ### Digital FAST: An AI-Driven Multimodal Framework for Rapid and Early Stroke Screening Authors: Ngoc-Khai Hoang, Thi-Nhu-Mai Nguyen, Huy-Hieu Pham Published: 2026-01-17 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.11896v2 Early identification of stroke symptoms is essential for enabling timely intervention and improving patient outcomes, particularly in prehospital settings. This study presents a fast, non-invasive multimodal deep learning framework for automatic binary stroke screening based on data collected during the F.A.S.T. assessment. The proposed approach integrates complementary information from facial expressions, speech signals, and upper-body movements to enhance diagnostic robustness. Facial dynamics are represented using landmark based features and modeled with a Transformer architecture to capture temporal dependencies. Speech signals are converted into mel spectrograms and processed using an Audio Spectrogram Transformer, while upper-body pose sequences are analyzed with an MLP-Mixer network to model spatiotemporal motion patterns. The extracted modality specific representations are combined through an attention-based fusion mechanism to effectively learn cross modal interactions. Experiments conducted on a self-collected dataset of 222 videos from 37 subjects demonstrate that the proposed multimodal model consistently outperforms unimodal baselines, achieving 95.83% accuracy and a 96.00% F1-score. The model attains a strong balance between sensitivity and specificity and successfully detects all stroke cases in the test set. These results highlight the potential of multimodal learning and transfer learning for early stroke screening, while emphasizing the need for larger, clinically representative datasets to support reliable real-world deployment. --- ### DevBench: A Realistic, Developer-Informed Benchmark for Code Generation Models Authors: Adarsh Kumarappan, Pareesa Ameneh Golnari, Wen Wen, Xiaoyu Liu, Gabriel Ryan, Yuting Sun, Shengyu Fu, Elsie Nallipogu Published: 2026-01-17 Categories: cs.LG, cs.AI, cs.SE arXiv: https://arxiv.org/abs/2601.11895v3 DevBench is a telemetry-driven benchmark designed to evaluate Large Language Models (LLMs) on realistic code completion tasks. It includes 1,800 evaluation instances across six programming languages and six task categories derived from real developer telemetry and synthesized using generator models from multiple provider families to mitigate single-source bias. Unlike prior benchmarks, it emphasizes ecological validity, avoids training data contamination, and enables detailed diagnostics. The evaluation combines functional correctness, similarity-based metrics, and LLM-judge assessments focused on usefulness and contextual relevance. 9 state-of-the-art models were assessed, with the strongest achieving only 43.5% Pass@1, confirming the benchmark remains challenging and revealing differences in syntactic precision, semantic reasoning, and practical utility. Our benchmark provides actionable insights to guide model selection and improvement, detail that is often missing from other benchmarks but is essential for both practical deployment and targeted model development. --- ### From Relative Entropy to Minimax: A Unified Framework for Coverage in MDPs Authors: Xihe Gu, Urbashi Mitra, Tara Javidi Published: 2026-01-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.11890v1 Targeted and deliberate exploration of state--action pairs is essential in reward-free Markov Decision Problems (MDPs). More precisely, different state-action pairs exhibit different degree of importance or difficulty which must be actively and explicitly built into a controlled exploration strategy. To this end, we propose a weighted and parameterized family of concave coverage objectives, denoted by $U_ρ$, defined directly over state--action occupancy measures. This family unifies several widely studied objectives within a single framework, including divergence-based marginal matching, weighted average coverage, and worst-case (minimax) coverage. While the concavity of $U_ρ$ captures the diminishing return associated with over-exploration, the simple closed form of the gradient of $U_ρ$ enables an explicit control to prioritize under-explored state--action pairs. Leveraging this structure, we develop a gradient-based algorithm that actively steers the induced occupancy toward a desired coverage pattern. Moreover, we show that as $ρ$ increases, the resulting exploration strategy increasingly emphasizes the least-explored state--action pairs, recovering worst-case coverage behavior in the limit. --- ### Agentic-R: Learning to Retrieve for Agentic Search Authors: Wenhan Liu, Xinyu Ma, Yutao Zhu, Yuchen Li, Daiting Shi, Dawei Yin, Zhicheng Dou Published: 2026-01-17 Categories: cs.IR, cs.CL arXiv: https://arxiv.org/abs/2601.11888v1 Agentic search has recently emerged as a powerful paradigm, where an agent interleaves multi-step reasoning with on-demand retrieval to solve complex questions. Despite its success, how to design a retriever for agentic search remains largely underexplored. Existing search agents typically rely on similarity-based retrievers, while similar passages are not always useful for final answer generation. In this paper, we propose a novel retriever training framework tailored for agentic search. Unlike retrievers designed for single-turn retrieval-augmented generation (RAG) that only rely on local passage utility, we propose to use both local query-passage relevance and global answer correctness to measure passage utility in a multi-turn agentic search. We further introduce an iterative training strategy, where the search agent and the retriever are optimized bidirectionally and iteratively. Different from RAG retrievers that are only trained once with fixed questions, our retriever is continuously improved using evolving and higher-quality queries from the agent. Extensive experiments on seven single-hop and multi-hop QA benchmarks demonstrate that our retriever, termed \ours{}, consistently outperforms strong baselines across different search agents. Our codes are available at: https://github.com/8421BCD/Agentic-R. --- ### Faithfulness vs. Safety: Evaluating LLM Behavior Under Counterfactual Medical Evidence Authors: Kaijie Mo, Siddhartha Venkatayogi, Chantal Shaib, Ramez Kouzy, Wei Xu, Byron C. Wallace, Junyi Jessy Li Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11886v2 In high-stakes domains like medicine, it may be generally desirable for models to faithfully adhere to the context provided. But what happens if the context does not align with model priors or safety protocols? In this paper, we investigate how LLMs behave and reason when presented with counterfactual (or even adversarial) medical evidence. We first construct MedCounterFact, a counterfactual medical QA dataset that requires the models to answer clinical comparison questions (i.e., judge the efficacy of certain treatments, with evidence consisting of randomized controlled trials provided as context). In MedCounterFact, real-world medical interventions within the questions and evidence are systematically replaced with four types of counterfactual stimuli, ranging from unknown words to toxic substances. Our evaluation across multiple frontier LLMs on MedCounterFact reveals that in the presence of counterfactual evidence, existing models overwhelmingly accept such "evidence" at face value even when it is dangerous or implausible, and provide confident and uncaveated answers. While it may be prudent to draw a boundary between faithfulness and safety, our findings suggest that models arguably overemphasize the former. --- ### MyGram: Modality-aware Graph Transformer with Global Distribution for Multi-modal Entity Alignment Authors: Zhifei Li, Ziyue Qin, Xiangyu Luo, Xiaoju Hou, Yue Zhao, Miao Zhang, Zhifang Huang, Kui Xiao, Bing Yang Published: 2026-01-17 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.11885v1 Multi-modal entity alignment aims to identify equivalent entities between two multi-modal Knowledge graphs by integrating multi-modal data, such as images and text, to enrich the semantic representations of entities. However, existing methods may overlook the structural contextual information within each modality, making them vulnerable to interference from shallow features. To address these challenges, we propose MyGram, a modality-aware graph transformer with global distribution for multi-modal entity alignment. Specifically, we develop a modality diffusion learning module to capture deep structural contextual information within modalities and enable fine-grained multi-modal fusion. In addition, we introduce a Gram Loss that acts as a regularization constraint by minimizing the volume of a 4-dimensional parallelotope formed by multi-modal features, thereby achieving global distribution consistency across modalities. We conduct experiments on five public datasets. Results show that MyGram outperforms baseline models, achieving a maximum improvement of 4.8% in Hits@1 on FBDB15K, 9.9% on FBYG15K, and 4.3% on DBP15K. --- ### Approximation Algorithm for Constrained $k$-Center Clustering: A Local Search Approach Authors: Chaoqi Jia, Longkun Guo, Kewen Liao, Zhigang Lu, Chao Chen, Jason Xue Published: 2026-01-17 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.11883v1 Clustering is a long-standing research problem and a fundamental tool in AI and data analysis. The traditional k-center problem, a fundamental theoretical challenge in clustering, has a best possible approximation ratio of 2, and any improvement to a ratio of 2 - ε would imply P = NP. In this work, we study the constrained k-center clustering problem, where instance-level cannot-link (CL) and must-link (ML) constraints are incorporated as background knowledge. Although general CL constraints significantly increase the hardness of approximation, previous work has shown that disjoint CL sets permit constant-factor approximations. However, whether local search can achieve such a guarantee in this setting remains an open question. To this end, we propose a novel local search framework based on a transformation to a dominating matching set problem, achieving the best possible approximation ratio of 2. The experimental results on both real-world and synthetic datasets demonstrate that our algorithm outperforms baselines in solution quality. --- ### TF-CoDiT: Conditional Time Series Synthesis with Diffusion Transformers for Treasury Futures Authors: Yingxiao Zhang, Jiaxin Duan, Junfu Zhang, Ke Feng Published: 2026-01-17 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.11880v1 Diffusion Transformers (DiT) have achieved milestones in synthesizing financial time-series data, such as stock prices and order flows. However, their performance in synthesizing treasury futures data is still underexplored. This work emphasizes the characteristics of treasury futures data, including its low volume, market dependencies, and the grouped correlations among multivariables. To overcome these challenges, we propose TF-CoDiT, the first DiT framework for language-controlled treasury futures synthesis. To facilitate low-data learning, TF-CoDiT adapts the standard DiT by transforming multi-channel 1-D time series into Discrete Wavelet Transform (DWT) coefficient matrices. A U-shape VAE is proposed to encode cross-channel dependencies hierarchically into a latent variable and bridge the latent and DWT spaces through decoding, thereby enabling latent diffusion generation. To derive prompts that cover essential conditions, we introduce the Financial Market Attribute Protocol (FinMAP) - a multi-level description system that standardizes daily$/$periodical market dynamics by recognizing 17$/$23 economic indicators from 7/8 perspectives. In our experiments, we gather four types of treasury futures data covering the period from 2015 to 2025, and define data synthesis tasks with durations ranging from one week to four months. Extensive evaluations demonstrate that TF-CoDiT can produce highly authentic data with errors at most 0.433 (MSE) and 0.453 (MAE) to the ground-truth. Further studies evidence the robustness of TF-CoDiT across contracts and temporal horizons. --- ### Accelerated MR Elastography Using Learned Neural Network Representation Authors: Xi Peng Published: 2026-01-17 Categories: eess.SP, cs.CV, cs.LG, q-bio.QM arXiv: https://arxiv.org/abs/2601.11878v2 To develop a deep-learning method for achieving fast high-resolution MR elastography from highly undersampled data without the need of high-quality training dataset. We first framed the deep neural network representation as a nonlinear extension of the linear subspace model, then used it to represent and reconstruct MRE image repetitions from undersampled k-space data. The network weights were learned using a multi-level k-space consistent loss. To further enhance reconstruction quality, phase-contrast specific magnitude and phase priors were incorporated, including the similarity of anatomical structures and smoothness of wave-induced harmonic displacement. Experiments were conducted using both 3D gradient-echo spiral and multi-slice spin-echo spiral MRE datasets. Compared to the conventional linear subspace-based approaches, the nonlinear network representation method was able to produce superior image reconstruction with suppressed noise and artifacts from a single in-plane spiral arm per MRE repetition (e.g., 2mm isotropic resolution in 1 min with a total R=10), yielding comparable stiffness estimation to the fully sampled data. This work demonstrated the feasibility of using deep network representations to model and reconstruct MRE images from highly-undersampled data, a nonlinear extension of the subspace-based approaches. --- ### AI for Green Spaces: Leveraging Autonomous Navigation and Computer Vision for Park Litter Removal Authors: Christopher Kao, Akhil Pathapati, James Davis Published: 2026-01-17 Categories: cs.RO, cs.AI, cs.CV, eess.SY arXiv: https://arxiv.org/abs/2601.11876v1 There are 50 billion pieces of litter in the U.S. alone. Grass fields contribute to this problem because picnickers tend to leave trash on the field. We propose building a robot that can autonomously navigate, identify, and pick up trash in parks. To autonomously navigate the park, we used a Spanning Tree Coverage (STC) algorithm to generate a coverage path the robot could follow. To navigate this path, we successfully used Real-Time Kinematic (RTK) GPS, which provides a centimeter-level reading every second. For computer vision, we utilized the ResNet50 Convolutional Neural Network (CNN), which detects trash with 94.52% accuracy. For trash pickup, we tested multiple design concepts. We select a new pickup mechanism that specifically targets the trash we encounter on the field. Our solution achieved an overall success rate of 80%, demonstrating that autonomous trash pickup robots on grass fields are a viable solution. --- ### GloCTM: Cross-Lingual Topic Modeling via a Global Context Space Authors: Nguyen Tien Phat, Ngo Vu Minh, Linh Van Ngo, Nguyen Thi Ngoc Diep, Thien Huu Nguyen Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11872v1 Cross-lingual topic modeling seeks to uncover coherent and semantically aligned topics across languages - a task central to multilingual understanding. Yet most existing models learn topics in disjoint, language-specific spaces and rely on alignment mechanisms (e.g., bilingual dictionaries) that often fail to capture deep cross-lingual semantics, resulting in loosely connected topic spaces. Moreover, these approaches often overlook the rich semantic signals embedded in multilingual pretrained representations, further limiting their ability to capture fine-grained alignment. We introduce GloCTM (Global Context Space for Cross-Lingual Topic Model), a novel framework that enforces cross-lingual topic alignment through a unified semantic space spanning the entire model pipeline. GloCTM constructs enriched input representations by expanding bag-of-words with cross-lingual lexical neighborhoods, and infers topic proportions using both local and global encoders, with their latent representations aligned through internal regularization. At the output level, the global topic-word distribution, defined over the combined vocabulary, structurally synchronizes topic meanings across languages. To further ground topics in deep semantic space, GloCTM incorporates a Centered Kernel Alignment (CKA) loss that aligns the latent topic space with multilingual contextual embeddings. Experiments across multiple benchmarks demonstrate that GloCTM significantly improves topic coherence and cross-lingual alignment, outperforming strong baselines. --- ### Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces Authors: Mike A. Merrill, Alexander G. Shaw, Nicholas Carlini, Boxuan Li, Harsh Raj, Ivan Bercovich, Lin Shi, Jeong Yeon Shin, Thomas Walshe, E. Kelly Buchanan, Junhong Shen, Guanghao Ye et al. Published: 2026-01-17 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2601.11868v1 AI agents may soon become capable of autonomously completing valuable, long-horizon tasks in diverse domains. Current benchmarks either do not measure real-world tasks, or are not sufficiently difficult to meaningfully measure frontier models. To this end, we present Terminal-Bench 2.0: a carefully curated hard benchmark composed of 89 tasks in computer terminal environments inspired by problems from real workflows. Each task features a unique environment, human-written solution, and comprehensive tests for verification. We show that frontier models and agents score less than 65\% on the benchmark and conduct an error analysis to identify areas for model and agent improvement. We publish the dataset and evaluation harness to assist developers and researchers in future work at https://www.tbench.ai/ . --- ### Advances in LLM Reasoning Enable Flexibility in Clinical Problem-Solving Authors: Kie Shidara, Preethi Prem, Jonathan Kim, Anna Podlasek, Feng Liu, Ahmed Alaa, Danilo Bernardo Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11866v1 Large Language Models (LLMs) have achieved high accuracy on medical question-answer (QA) benchmarks, yet their capacity for flexible clinical reasoning has been debated. Here, we asked whether advances in reasoning LLMs improve their cognitive flexibility in clinical reasoning. We assessed reasoning models from the OpenAI, Grok, Gemini, Claude, and DeepSeek families on the medicine abstraction and reasoning corpus (mARC), an adversarial medical QA benchmark which utilizes the Einstellung effect to induce inflexible overreliance on learned heuristic patterns in contexts where they become suboptimal. We found that strong reasoning models avoided Einstellung-based traps more often than weaker reasoning models, achieving human-level performance on mARC. On questions most commonly missed by physicians, the top 5 performing models answered 55% to 70% correctly with high confidence, indicating that these models may be less susceptible than humans to Einstellung effects. Our results indicate that strong reasoning models demonstrate improved flexibility in medical reasoning, achieving performance on par with humans on mARC. --- ### CTPD: Cross Tokenizer Preference Distillation Authors: Truong Nguyen, Phi Van Dat, Ngan Nguyen, Linh Ngo Van, Trung Le, Thanh Hong Nguyen Published: 2026-01-17 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.11865v1 While knowledge distillation has seen widespread use in pre-training and instruction tuning, its application to aligning language models with human preferences remains underexplored, particularly in the more realistic cross-tokenizer setting. The incompatibility of tokenization schemes between teacher and student models has largely prevented fine-grained, white-box distillation of preference information. To address this gap, we propose Cross-Tokenizer Preference Distillation (CTPD), the first unified framework for transferring human-aligned behavior between models with heterogeneous tokenizers. CTPD introduces three key innovations: (1) Aligned Span Projection, which maps teacher and student tokens to shared character-level spans for precise supervision transfer; (2) a cross-tokenizer adaptation of Token-level Importance Sampling (TIS-DPO) for improved credit assignment; and (3) a Teacher-Anchored Reference, allowing the student to directly leverage the teacher's preferences in a DPO-style objective. Our theoretical analysis grounds CTPD in importance sampling, and experiments across multiple benchmarks confirm its effectiveness, with significant performance gains over existing methods. These results establish CTPD as a practical and general solution for preference distillation across diverse tokenization schemes, opening the door to more accessible and efficient alignment of language models. --- ### AGGC: Adaptive Group Gradient Clipping for Stabilizing Large Language Model Training Authors: Zhiyuan Li, Yuan Wu, Yi Chang Published: 2026-01-17 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2601.11864v1 To stabilize the training of Large Language Models (LLMs), gradient clipping is a nearly ubiquitous heuristic used to alleviate exploding gradients. However, traditional global norm clipping erroneously presupposes gradient homogeneity across different functional modules, leading to an adverse "spill-over" effect where volatile parameters force unnecessary scaling on stable ones. To overcome this, we propose Adaptive Group-wise Gradient Clipping (AGGC). AGGC partitions parameters into groups based on functional types and regulates each according to its historical behavior using an Exponential Moving Average (EMA). Specifically, it constructs an adaptive interval to simultaneously mitigate gradient explosion and vanishing, while employing a time-dependent scheduling mechanism to balance exploration and convergence. Experiments on LLaMA 2-7B, Mistral-7B, and Gemma-7B models show that AGGC consistently outperforms LoRA and frequently surpasses Full Fine-Tuning. On the GSM8K benchmark, Mistral-7B fine-tuned with AGGC achieves an accuracy of 72.93%, exceeding LoRA's 69.5%. AGGC also effectively stabilizes Reinforcement Learning with Verifiable Rewards (RLVR), enhancing the logic deduction of Qwen 2.5 and Llama 3.2 models. Experimental results demonstrate that AGGC effectively addresses the limitations of traditional gradient clipping methods, particularly in overcoming gradient heterogeneity, by utilizing a modular, adaptive clipping strategy to stabilize the training process. Due to its lightweight design, AGGC can be seamlessly integrated into existing post-training pipelines with negligible overhead. ---