# Research Papers — 2026-01 · bundle page 32 of 189

> Full-text markdown of papers #1551–1600 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/32
[← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/31.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-01/page/33.md)

### A New Mode of Teaching Chinese as a Foreign Language from the Perspective of Smart System Studied by Using Rongzhixue

Authors: Xiaohui Zou, Lijun Ke, Shunpeng Zou
Published: 2026-01-28
Categories: cs.CY, cs.AI, cs.HC
arXiv: https://arxiv.org/abs/2602.06992v1

The purpose of this study is to introduce a new model of teaching Chinese as a foreign language from the perspective of integrating wisdom. Its characteristics are as follows: focusing on the butterfly model of interpretation before translation, highlighting the new method of bilingual thinking training, on the one hand, applying the new theory of Chinese characters, the theory of the relationship between language and speech, and the forward-looking research results of language science; On the other hand, the application of the new model of teaching Chinese as a foreign language, AI empowering teaching and learning, and the forward-looking research results of educational science fully reflect a series of characteristics of the new model of teaching Chinese as a foreign language from the perspective of integrating wisdom. Its beneficial effects are: not only the old view of language and education, especially the old view of teaching Chinese as a foreign language, but also the old view of human-computer interaction. Its significance lies in that a series of great cross-border Rongzhixue such as language, knowledge, education and teaching, as well as new methods and new topics of bilingual thinking training are clearly put forward from the perspective of integrating wisdom. Especially in the face of the challenge of Chat GPT to human learning ability and even creativity, the existing concepts of language knowledge education and teaching are already very backward. The old concepts of Chinese language education, and teaching Chinese as a foreign language are all facing a series of subversive innovation challenges. How to seek changes in adaptation? This study has made a series of innovative attempts, hoping to benefit academic colleagues, teachers and students.

---

### LangGS-SLAM: Real-Time Language-Feature Gaussian Splatting SLAM

Authors: Seongbo Ha, Sibaek Lee, Kyungsu Kang, Joonyeol Choi, Seungjun Tak, Hyeonwoo Yu
Published: 2026-01-28
Categories: cs.RO, cs.CV, eess.IV
arXiv: https://arxiv.org/abs/2602.06991v1

In this paper, we propose a RGB-D SLAM system that reconstructs a language-aligned dense feature field while sustaining low-latency tracking and mapping. First, we introduce a Top-K Rendering pipeline, a high-throughput and semantic-distortion-free method for efficiently rendering high-dimensional feature maps. To address the resulting semantic-geometric discrepancy and mitigate the memory consumption, we further design a multi-criteria map management strategy that prunes redundant or inconsistent Gaussians while preserving scene integrity. Finally, a hybrid field optimization framework jointly refines the geometric and semantic fields under real-time constraints by decoupling their optimization frequencies according to field characteristics. The proposed system achieves superior geometric fidelity compared to geometric-only baselines and comparable semantic fidelity to offline approaches while operating at 15 FPS. Our results demonstrate that online SLAM with dense, uncompressed language-aligned feature fields is both feasible and effective, bridging the gap between 3D perception and language-based reasoning.

---

### How Much Information Can a Vision Token Hold? A Scaling Law for Recognition Limits in VLMs

Authors: Shuxin Zhuang, Zi Liang, Runsheng Yu, Hongzong Li, Rong Feng, Shiqin Tang, Youzhi Zhang
Published: 2026-01-28
Categories: cs.LG, cs.CV
arXiv: https://arxiv.org/abs/2602.02539v1

Recent vision-centric approaches have made significant strides in long-context modeling. Represented by DeepSeek-OCR, these models encode rendered text into continuous vision tokens, achieving high compression rates without sacrificing recognition precision. However, viewing the vision encoder as a lossy channel with finite representational capacity raises a fundamental question: what is the information upper bound of visual tokens? To investigate this limit, we conduct controlled stress tests by progressively increasing the information quantity (character count) within an image. We observe a distinct phase-transition phenomenon characterized by three regimes: a near-perfect Stable Phase, an Instability Phase marked by increased error variance, and a total Collapse Phase. We analyze the mechanical origins of these transitions and identify key factors. Furthermore, we formulate a probabilistic scaling law that unifies average vision token load and visual density into a latent difficulty metric. Extensive experiments across various Vision-Language Models demonstrate the universality of this scaling law, providing critical empirical guidance for optimizing the efficiency-accuracy trade-off in visual context compression.

---

### Enhancing Post-Training Quantization via Future Activation Awareness

Authors: Zheqi Lv, Zhenxuan Fan, Qi Tian, Wenqiao Zhang, Yueting Zhuang
Published: 2026-01-28
Categories: cs.LG, cs.CL, cs.CV
arXiv: https://arxiv.org/abs/2602.02538v1

Post-training quantization (PTQ) is a widely used method to compress large language models (LLMs) without fine-tuning. It typically sets quantization hyperparameters (e.g., scaling factors) based on current-layer activations. Although this method is efficient, it suffers from quantization bias and error accumulation, resulting in suboptimal and unstable quantization, especially when the calibration data is biased. To overcome these issues, we propose Future-Aware Quantization (FAQ), which leverages future-layer activations to guide quantization. This allows better identification and preservation of important weights, while reducing sensitivity to calibration noise. We further introduce a window-wise preview mechanism to softly aggregate multiple future-layer activations, mitigating over-reliance on any single layer. To avoid expensive greedy search, we use a pre-searched configuration to minimize overhead. Experiments show that FAQ consistently outperforms prior methods with negligible extra cost, requiring no backward passes, data reconstruction, or tuning, making it well-suited for edge deployment.

---

### WorldVQA: Measuring Atomic World Knowledge in Multimodal Large Language Models

Authors: Runjie Zhou, Youbo Shao, Haoyu Lu, Bowei Xing, Tongtong Bai, Yujie Chen, Jie Zhao, Lin Sui, Haotian Yao, Zijia Zhao, Hao Yang, Haoning Wu et al.
Published: 2026-01-28
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2602.02537v1

We introduce WorldVQA, a benchmark designed to evaluate the atomic visual world knowledge of Multimodal Large Language Models (MLLMs). Unlike current evaluations, which often conflate visual knowledge retrieval with reasoning, WorldVQA decouples these capabilities to strictly measure "what the model memorizes." The benchmark assesses the atomic capability of grounding and naming visual entities across a stratified taxonomy, spanning from common head-class objects to long-tail rarities. We expect WorldVQA to serve as a rigorous test for visual factuality, thereby establishing a standard for assessing the encyclopedic breadth and hallucination rates of current and next-generation frontier models.

---

### From Sparse Decisions to Dense Reasoning: A Multi-attribute Trajectory Paradigm for Multimodal Moderation

Authors: Tianle Gu, Kexin Huang, Lingyu Li, Ruilin Luo, Shiyang Huang, Zongqi Wang, Yujiu Yang, Yan Teng, Yingchun Wang
Published: 2026-01-28
Categories: cs.LG, cs.AI, cs.CL, cs.CV
arXiv: https://arxiv.org/abs/2602.02536v1

Safety moderation is pivotal for identifying harmful content. Despite the success of textual safety moderation, its multimodal counterparts remain hindered by a dual sparsity of data and supervision. Conventional reliance on binary labels lead to shortcut learning, which obscures the intrinsic classification boundaries necessary for effective multimodal discrimination. Hence, we propose a novel learning paradigm (UniMod) that transitions from sparse decision-making to dense reasoning traces. By constructing structured trajectories encompassing evidence grounding, modality assessment, risk mapping, policy decision, and response generation, we reformulate monolithic decision tasks into a multi-dimensional boundary learning process. This approach forces the model to ground its decision in explicit safety semantics, preventing the model from converging on superficial shortcuts. To facilitate this paradigm, we develop a multi-head scalar reward model (UniRM). UniRM provides multi-dimensional supervision by assigning attribute-level scores to the response generation stage. Furthermore, we introduce specialized optimization strategies to decouple task-specific parameters and rebalance training dynamics, effectively resolving interference between diverse objectives in multi-task learning. Empirical results show UniMod achieves competitive textual moderation performance and sets a new multimodal benchmark using less than 40\% of the training data used by leading baselines. Ablations further validate our multi-attribute trajectory reasoning, offering an effective and efficient framework for multimodal moderation. Supplementary materials are available at \href{https://trustworthylab.github.io/UniMod/}{project website}.

---

### Enhancing Psychologists' Understanding through Explainable Deep Learning Framework for ADHD Diagnosis

Authors: Abdul Rehman, Ilona Heldal, Jerry Chun-Wei Lin
Published: 2026-01-28
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2602.02535v1

Attention Deficit Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder that is challenging to diagnose and requires advanced approaches for reliable and transparent identification and classification. It is characterized by a pattern of inattention, hyperactivity and impulsivity that is more severe and more frequent than in individuals with a comparable level of development. In this paper, an explainable framework based on a fine-tuned hybrid Deep Neural Network (DNN) and Recurrent Neural Network (RNN) called HyExDNN-RNN model is proposed for ADHD detection, multi-class categorization, and decision interpretation. This framework not only detects ADHD, but also provides interpretable insights into the diagnostic process so that psychologists can better understand and trust the results of the diagnosis. We use the Pearson correlation coefficient for optimal feature selection and machine and deep learning models for experimental analysis and comparison. We use a standardized technique for feature reduction, model selection and interpretation to accurately determine the diagnosis rate and ensure the interpretability of the proposed framework. Our framework provided excellent results on binary classification, with HyExDNN-RNN achieving an F1 score of 99% and 94.2% on multi-class categorization. XAI approaches, in particular SHapley Additive exPlanations (SHAP) and Permutation Feature Importance (PFI), provided important insights into the importance of features and the decision logic of models. By combining AI with human expertise, we aim to bridge the gap between advanced computational techniques and practical psychological applications. These results demonstrate the potential of our framework to assist in ADHD diagnosis and interpretation.

---

### HMVLA: Hyperbolic Multimodal Fusion for Vision-Language-Action Models

Authors: Kun Wang, Xiao Feng, Mingcheng Qu, Tonghua Su
Published: 2026-01-28
Categories: cs.RO, cs.LG
arXiv: https://arxiv.org/abs/2602.02533v1

Vision Language Action (VLA) models have recently shown great potential in bridging multimodal perception with robotic control. However, existing methods often rely on direct fine-tuning of pre-trained Vision-Language Models (VLMs), feeding semantic and visual features directly into a policy network without fully addressing the unique semantic alignment challenges in the VLA domain. In this paper, we propose HMVLA, a novel VLA framework that exploits the inherent hierarchical structures in vision and language for comprehensive semantic alignment. Unlike traditional methods that perform alignment in Euclidean space, our HMVLA embeds multimodal features in hyperbolic space, enabling more effective modeling of the hierarchical relationships present in image text data. Furthermore, we introduce a sparsely gated Mixture of Experts (MoE) mechanism tailored for semantic alignment, which enhances multimodal comprehension between images and text while improving efficiency. Extensive experiments demonstrate that HMVLA surpasses baseline methods in both accuracy and generalization. In addition, we validate its robustness by reconstructing datasets to further test cross domain adaptability.

---

### CADENT: Gated Hybrid Distillation for Sample-Efficient Transfer in Reinforcement Learning

Authors: Mahyar Alinejad, Yue Wang, George Atia
Published: 2026-01-28
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2602.02532v1

Transfer learning promises to reduce the high sample complexity of deep reinforcement learning (RL), yet existing methods struggle with domain shift between source and target environments. Policy distillation provides powerful tactical guidance but fails to transfer long-term strategic knowledge, while automaton-based methods capture task structure but lack fine-grained action guidance. This paper introduces Context-Aware Distillation with Experience-gated Transfer (CADENT), a framework that unifies strategic automaton-based knowledge with tactical policy-level knowledge into a coherent guidance signal. CADENT's key innovation is an experience-gated trust mechanism that dynamically weighs teacher guidance against the student's own experience at the state-action level, enabling graceful adaptation to target domain specifics. Across challenging environments, from sparse-reward grid worlds to continuous control tasks, CADENT achieves 40-60\% better sample efficiency than baselines while maintaining superior asymptotic performance, establishing a robust approach for adaptive knowledge transfer in RL.

---

### Toward a Unified Semantic Loss Model for Deep JSCC-based Transmission of EO Imagery

Authors: Ti Ti Nguyen, Thanh-Dung Le, Vu Nguyen Ha, Duc-Dung Tran, Hung Nguyen-Kha, Dinh-Hieu Tran, Carlos L. Marcos-Rojas, Juan C. Merlano-Duncan, Symeon Chatzinotas
Published: 2026-01-28
Categories: eess.IV, cs.CV
arXiv: https://arxiv.org/abs/2602.00136v1

Modern Earth Observation (EO) systems increasingly rely on high-resolution imagery to support critical applications such as environmental monitoring, disaster response, and land-use analysis. Although these applications benefit from detailed visual data, the resulting data volumes impose significant challenges on satellite communication systems constrained by limited bandwidth, power, and dynamic link conditions. To address these limitations, this paper investigates Deep Joint Source-Channel Coding (DJSCC) as an effective source-channel paradigm for the transmission of EO imagery. We focus on two complementary aspects of semantic loss in DJSCC-based systems. First, a reconstruction-centric framework is evaluated by analyzing the semantic degradation of reconstructed images under varying compression ratios and channel signal-to-noise ratios (SNR). Second, a task-oriented framework is developed by integrating DJSCC with lightweight, application-specific models (e.g., EfficientViT), with performance measured using downstream task accuracy rather than pixel-level fidelity. Based on extensive empirical analysis, we propose a unified semantic loss framework that captures both reconstruction-centric and task-oriented performance within a single model. This framework characterizes the implicit relationship between JSCC compression, channel SNR, and semantic quality, offering actionable insights for the design of robust and efficient EO imagery transmission under resource-constrained satellite links.

---

### LLaVA-FA: Learning Fourier Approximation for Compressing Large Multimodal Models

Authors: Pengcheng Zheng, Chaoning Zhang, Jiarong Mo, GuoHui Li, Jiaquan Zhang, Jiahao Zhang, Sihan Cao, Sheng Zheng, Caiyan Qin, Guoqing Wang, Yang Yang
Published: 2026-01-28
Categories: cs.CV
arXiv: https://arxiv.org/abs/2602.00135v1

Large multimodal models (LMMs) have achieved impressive performance on various vision-language tasks, but their substantial computational and memory costs hinder their practical deployment. Existing compression methods often decouple low-rank decomposition and quantization, leading to compounded reconstruction errors, especially in multimodal architectures with cross-modal redundancy. To address this issue, we propose LLaVA-FA, a novel efficient LMM that performs joint low-rank plus quantization approximation in the frequency domain. By leveraging the de-correlation and conjugate symmetry properties of Fourier transform, LLaVA-FA achieves more compact and accurate weight representations. Furthermore, we introduce PolarQuant, a polar-coordinate quantization method tailored for complex matrices, and an optional diagonal calibration (ODC) scheme that eliminates the need for large-scale calibration data. Extensive experimental results demonstrate that our proposed LLaVA-FA outperforms existing efficient multimodal models across multiple benchmarks while maintaining minimal activated parameters and low computational costs, validating its effectiveness as a powerful solution for compressing LMMs.

---

### PredictionMarketBench: A SWE-bench-Style Framework for Backtesting Trading Agents on Prediction Markets

Authors: Avi Arora, Ritesh Malpani
Published: 2026-01-28
Categories: q-fin.ST, cs.AI
arXiv: https://arxiv.org/abs/2602.00133v1

Prediction markets offer a natural testbed for trading agents: contracts have binary payoffs, prices can be interpreted as probabilities, and realized performance depends critically on market microstructure, fees, and settlement risk. We introduce PredictionMarketBench, a SWE-bench-style benchmark for evaluating algorithmic and LLM-based trading agents on prediction markets via deterministic, event-driven replay of historical limit-order-book and trade data. PredictionMarketBench standardizes (i) episode construction from raw exchange streams (orderbooks, trades, lifecycle, settlement), (ii) an execution-realistic simulator with maker/taker semantics and fee modeling, and (iii) a tool-based agent interface that supports both classical strategies and tool-calling LLM agents with reproducible trajectories. We release four Kalshi-based episodes spanning cryptocurrency, weather, and sports. Baseline results show that naive trading agents can underperform due to transaction costs and settlement losses, while fee-aware algorithmic strategies remain competitive in volatile episodes.

---

### Shedding the Facades, Connecting the Domains: Detecting Shifting Multimodal Hate Video with Test-Time Adaptation

Authors: Jiao Li, Jian Lang, Xikai Tang, Wenzheng Shu, Ting Zhong, Qiang Gao, Yong Wang, Leiting Chen, Fan Zhou
Published: 2026-01-28
Categories: cs.CV
arXiv: https://arxiv.org/abs/2602.00132v1

Hate Video Detection (HVD) is crucial for online ecosystems. Existing methods assume identical distributions between training (source) and inference (target) data. However, hateful content often evolves into irregular and ambiguous forms to evade censorship, resulting in substantial semantic drift and rendering previously trained models ineffective. Test-Time Adaptation (TTA) offers a solution by adapting models during inference to narrow the cross-domain gap, while conventional TTA methods target mild distribution shifts and struggle with the severe semantic drift in HVD. To tackle these challenges, we propose SCANNER, the first TTA framework tailored for HVD. Motivated by the insight that, despite the evolving nature of hateful manifestations, their underlying cores remain largely invariant (i.e., targeting is still based on characteristics like gender, race, etc), we leverage these stable cores as a bridge to connect the source and target domains. Specifically, SCANNER initially reveals the stable cores from the ambiguous layout in evolving hateful content via a principled centroid-guided alignment mechanism. To alleviate the impact of outlier-like samples that are weakly correlated with centroids during the alignment process, SCANNER enhances the prior by incorporating a sample-level adaptive centroid alignment strategy, promoting more stable adaptation. Furthermore, to mitigate semantic collapse from overly uniform outputs within clusters, SCANNER introduces an intra-cluster diversity regularization that encourages the cluster-wise semantic richness. Experiments show that SCANNER outperforms all baselines, with an average gain of 4.69% in Macro-F1 over the best.

---

### PovNet+: A Deep Learning Architecture for Socially Assistive Robots to Learn and Assist with Multiple Activities of Daily Living

Authors: Fraser Robinson, Souren Pashangpour, Matthew Lisondra, Goldie Nejat
Published: 2026-01-28
Categories: cs.CV, cs.RO
arXiv: https://arxiv.org/abs/2602.00131v1

A significant barrier to the long-term deployment of autonomous socially assistive robots is their inability to both perceive and assist with multiple activities of daily living (ADLs). In this paper, we present the first multimodal deep learning architecture, POVNet+, for multi-activity recognition for socially assistive robots to proactively initiate assistive behaviors. Our novel architecture introduces the use of both ADL and motion embedding spaces to uniquely distinguish between a known ADL being performed, a new unseen ADL, or a known ADL being performed atypically in order to assist people in real scenarios. Furthermore, we apply a novel user state estimation method to the motion embedding space to recognize new ADLs while monitoring user performance. This ADL perception information is used to proactively initiate robot assistive interactions. Comparison experiments with state-of-the-art human activity recognition methods show our POVNet+ method has higher ADL classification accuracy. Human-robot interaction experiments in a cluttered living environment with multiple users and the socially assistive robot Leia using POVNet+ demonstrate the ability of our multi-modal ADL architecture in successfully identifying different seen and unseen ADLs, and ADLs being performed atypically, while initiating appropriate assistive human-robot interactions.

---

### On the Relationship Between Representation Geometry and Generalization in Deep Neural Networks

Authors: Sumit Yadav
Published: 2026-01-28
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.00130v2

We investigate the relationship between representation geometry and neural network performance. Analyzing 52 pretrained ImageNet models across 13 architecture families, we show that effective dimension -- an unsupervised geometric metric -- strongly predicts accuracy. Output effective dimension achieves partial r=0.75 ($p < 10^(-10)$) after controlling for model capacity, while total compression achieves partial r=-0.72. These findings replicate across ImageNet and CIFAR-10, and generalize to NLP: effective dimension predicts performance for 8 encoder models on SST-2/MNLI and 15 decoder-only LLMs on AG News (r=0.69, p=0.004), while model size does not (r=0.07). We establish bidirectional causality: degrading geometry via noise causes accuracy loss (r=-0.94, $p < 10^(-9)$), while improving geometry via PCA maintains accuracy across architectures (-0.03pp at 95% variance). This relationship is noise-type agnostic -- Gaussian, Uniform, Dropout, and Salt-and-pepper noise all show $|r| > 0.90$. These results establish that effective dimension provides domain-agnostic predictive and causal information about neural network performance, computed entirely without labels.

---

### Monte Carlo Tree Search for Execution-Guided Program Repair with Large Language Models

Authors: Yixuan Liang
Published: 2026-01-28
Categories: cs.LG, cs.SE
arXiv: https://arxiv.org/abs/2602.00129v1

Automated program repair with large language models remains challenging at the repository level due to long-horizon reasoning requirements and the limitations of autoregressive decoding. We present CodePilot, a hybrid framework that integrates Monte Carlo Tree Search (MCTS) with large language models to enable execution-guided program repair for real-world GitHub issues. CodePilot performs hierarchical fault localization from repository to file and function level, explores diverse patch trajectories using MCTS, and leverages execution feedback as a reward signal to guide search and refinement. The framework further incorporates confidence-calibrated generation to selectively refine low-confidence outputs. Experiments on SWE-bench Lite demonstrate that CodePilot achieves a 24.67% issue resolution rate using open-weight models, outperforming comparable baselines. These results suggest that combining symbolic search with neural language models is an effective strategy for scalable, execution-aware software engineering automation.

---

### Quantum Model Parallelism for MRI-Based Classification of Alzheimer's Disease Stages

Authors: Emine Akpinar, Murat Oduncuoglu
Published: 2026-01-28
Categories: cs.LG, quant-ph
arXiv: https://arxiv.org/abs/2602.00128v2

With increasing life expectancy, AD has become a major global health concern. While classical AI-based methods have been developed for early diagnosis and stage classification of AD, growing data volumes and limited computational resources necessitate faster, more efficient approaches. Quantum-based AI methods, which leverage superposition and entanglement principles along with high-dimensional Hilbert space, can surpass classical approaches' limitations and offer higher accuracy for high-dimensional, heterogeneous, and noisy data. In this study, a Quantum-Based Parallel Model (QBPM) architecture is proposed for the efficient classification of AD stages using MRI datasets, inspired by the principles of classical model parallelism. The proposed model leverages quantum advantages by employing two distinct quantum circuits, each incorporating rotational and entanglement blocks, running in parallel on the same quantum simulator. The classification performance of the model was evaluated on two different datasets to assess its overall robustness and generalization capability. The proposed model demonstrated high classification accuracy across both datasets, highlighting its overall robustness and generalization capability. Results obtained under high-level Gaussian noise, simulating real-world conditions, further provided experimental evidence for the model's applicability not only in theoretical but also in practical scenarios. Moreover, compared with five different classical transfer learning methods, the proposed model demonstrated its efficiency as an alternative to classical approaches by achieving higher classification accuracy and comparable execution time while utilizing fewer circuit parameters. The results indicate that the proposed QBPM architecture represents an innovative and powerful approach for the classification of stages in complex diseases such as Alzheimer's.

---

### ALIGN: Aligned Delegation with Performance Guarantees for Multi-Agent LLM Reasoning

Authors: Tong Zhu, Baiting Chen, Jin Zhou, Hua Zhou, Sriram Sankararaman, Xiaowu Dai
Published: 2026-01-28
Categories: cs.LG
arXiv: https://arxiv.org/abs/2602.00127v1

LLMs often underperform on complex reasoning tasks when relying on a single generation-and-selection pipeline. Inference-time ensemble methods can improve performance by sampling diverse reasoning paths or aggregating multiple candidate answers, but they typically treat candidates independently and provide no formal guarantees that ensembling improves reasoning quality. We propose a novel method, Aligned Delegation for Multi-Agent LLM Reasoning (ALIGN), which formulates LLM reasoning as an aligned delegation game. In ALIGN, a principal delegates a task to multiple agents that generate candidate solutions under designed incentives, and then selects among their outputs to produce a final answer. This formulation induces structured interaction among agents while preserving alignment between agent and principal objectives. We establish theoretical guarantees showing that, under a fair comparison with equal access to candidate solutions, ALIGN provably improves expected performance over single-agent generation. Our analysis accommodates correlated candidate answers and relaxes independence assumptions that are commonly used in prior work. Empirical results across a broad range of LLM reasoning benchmarks consistently demonstrate that ALIGN outperforms strong single-agent and ensemble baselines.

---

### Tacit Coordination of Large Language Models

Authors: Ido Aharon, Emanuele La Malfa, Michael Wooldridge, Sarit Kraus
Published: 2026-01-28
Categories: cs.GT, cs.LG, cs.MA
arXiv: https://arxiv.org/abs/2601.22184v2

Large Language Models (LLMs) are increasingly deployed in multi-agent settings that require coordination without communication, from human-AI interaction to safety-critical scenarios. Humans often overcome the absence of communication through focal points: salient solutions that naturally stand out to all participants. We present the first large-scale evaluation of how, when, and why focal points emerge in LLMs, comparing their behaviour with humans across cooperative and competitive games, including realistic search and rescue scenarios, demonstrating when focal points enable effective coordination. Across more than 20 open- and closed-source models, we find that LLMs exhibit a remarkable ability to coordinate without communication, often matching or outperforming humans. However, the same models consistently fail in tasks requiring numerical common sense or culturally nuanced notions of salience. We additionally evaluate simple learning-free strategies that substantially improve coordination both among LLMs and between humans and LLMs. Our results reveal striking coordination capabilities, as well as social limitations in modern LLMs, and offer new insight into the latent notions of salience encoded within them. Our findings caution against assuming that LLMs share humans' cultural and perceptual substrate when deployed in coordination settings.

---

### COL-Trees: Efficient Hierarchical Object Search in Road Networks

Authors: Tenindra Abeywickrama, Muhammad Aamir Cheema, Sabine Storandt
Published: 2026-01-28
Categories: cs.DB, cs.AI, cs.DS
arXiv: https://arxiv.org/abs/2601.22183v1

Location-based services rely heavily on efficient methods that search for relevant points-of-interest (POIs) near a given location. A k Nearest Neighbor (kNN) query is one such example that finds the k closest POIs from an agent's location. While most existing techniques focus on retrieving nearby POIs for a single agent, these search heuristics do not translate to many other applications. For example, Aggregate k Nearest Neighbor (AkNN) queries require POIs that are close to multiple agents. k Farthest Neighbor (kFN) queries require POIs that are the antithesis of nearest. Such problems naturally benefit from a hierarchical approach, but existing methods rely on Euclidean-based heuristics, which have diminished effectiveness in graphs such as road networks. We propose a novel data structure, COL-Tree (Compacted Object-Landmark Tree), to address this gap by enabling efficient hierarchical graph traversal using a more accurate landmark-based heuristic. We then present query algorithms that utilize COL-Trees to efficiently answer AkNN, kFN, and other queries. In our experiments on real-world and synthetic datasets, we demonstrate that our techniques significantly outperform existing approaches, achieving up to 4 orders of magnitude improvement. Moreover, this comes at a small pre-processing overhead in both theory and practice.

---

### ShellForge: Adversarial Co-Evolution of Webshell Generation and Multi-View Detection for Robust Webshell Defense

Authors: Yizhong Ding
Published: 2026-01-28
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2601.22182v1

Webshells remain a primary foothold for attackers to compromise servers, particularly within PHP ecosystems. However, existing detection mechanisms often struggle to keep pace with rapid variant evolution and sophisticated obfuscation techniques that camouflage malicious intent. Furthermore, many current defenses suffer from high false-alarm rates when encountering benign administrative scripts that employ heavy obfuscation for intellectual property protection. To address these challenges, we present ShellForge, an adversarial co-evolution framework that couples automated webshell generation with multi-view detection to continuously harden defensive boundaries. The framework operates through an iterative co-training loop where a generator and a detector mutually reinforce each other via the exchange of hard samples. The generator is optimized through supervised fine-tuning and preference-based reinforcement learning to synthesize functional, highly evasive variants. Simultaneously, we develop a multi-view fusion detector that integrates semantic features from long-string compression, structural features from pruned abstract syntax trees, and global statistical indicators such as Shannon entropy. To minimize false positives, ShellForge utilizes a LLM-based transformation to create de-malicious samples--scripts that retain complex obfuscation patterns but lack harmful payloads--serving as high-quality hard negatives during training. Evaluations on the public FWOID benchmark demonstrate that ShellForge significantly enhances defensive robustness. Upon convergence, the detector maintains a 0.981 F1-score while the generator achieves a 0.939 evasion rate against commercial engines on VirusTotal.

---

### MrRoPE: Mixed-radix Rotary Position Embedding

Authors: Qingyuan Tian, Wenhong Zhu, Xiaoran Liu, Xiaofeng Wang, Rui Wang
Published: 2026-01-28
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.22181v1

Rotary Position Embedding (RoPE)-extension refers to modifying or generalizing the Rotary Position Embedding scheme to handle longer sequences than those encountered during pre-training. However, current extension strategies are highly diverse and lack a unified theoretical foundation. In this paper, we propose MrRoPE (Mixed-radix RoPE), a generalized encoding formulation based on a radix system conversion perspective, which elegantly unifies various RoPE-extension approaches as distinct radix conversion strategies. Based on this theory, we introduce two training-free extensions, MrRoPE-Uni and MrRoPE-Pro, which leverage uniform and progressive radix conversion strategies, respectively, to achieve 'train short, test long' generalization. Without fine-tuning, MrRoPE-Pro sustains over 85% recall in the 128K-context Needle-in-a-Haystack test and achieves more than double YaRN's accuracy on Infinite-Bench retrieval and dialogue subsets. Theoretical analysis confirms that MrRoPE-Pro effectively raises the upper bound of RoPE's attainable encoding length, which further validates the reliability and utility of our theory and methodology.

---

### PhaseCoder: Microphone Geometry-Agnostic Spatial Audio Understanding for Multimodal LLMs

Authors: Artem Dementyev, Wazeer Zulfikar, Sinan Hersek, Pascal Getreuer, Anurag Kumar, Vivek Kumar
Published: 2026-01-28
Categories: cs.SD, cs.AI, eess.AS
arXiv: https://arxiv.org/abs/2601.21124v2

Current multimodal LLMs process audio as a mono stream, ignoring the rich spatial information essential for embodied AI. Existing spatial audio models, conversely, are constrained to fixed microphone geometries, preventing deployment across diverse devices. We present PhaseCoder, a transformer-only spatial audio encoder that is agnostic to microphone geometry. PhaseCoder takes raw multichannel audio and microphone coordinates as inputs to perform localization and produces robust spatial embeddings. We demonstrate that Gemma 3n LLM can be fine-tuned to reason over "Spatial Audio Tokens" produced by PhaseCoder. We show our encoder achieves state-of-the-art results on microphone-invariant localization benchmarks and, for the first time, enables an LLM to perform complex spatial reasoning and targeted transcription tasks from an arbitrary microphone array.

---

### CUA-Skill: Develop Skills for Computer Using Agent

Authors: Tianyi Chen, Yinheng Li, Michael Solodko, Sen Wang, Nan Jiang, Tingyuan Cui, Junheng Hao, Jongwoo Ko, Sara Abdali, Leon Xu, Suzhen Zheng, Hao Fan et al.
Published: 2026-01-28
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21123v2

Computer-Using Agents (CUAs) aim to autonomously operate computer systems to complete real-world tasks. However, existing agentic systems remain difficult to scale and lag behind human performance. A key limitation is the absence of reusable and structured skill abstractions that capture how humans interact with graphical user interfaces and how to leverage these skills. We introduce CUA-Skill, a computer-using agentic skill base that encodes human computer-use knowledge as skills coupled with parameterized execution and composition graphs. CUA-Skill is a large-scale library of carefully engineered skills spanning common Windows applications, serving as a practical infrastructure and tool substrate for scalable, reliable agent development. Built upon this skill base, we construct CUA-Skill Agent, an end-to-end computer-using agent that supports dynamic skill retrieval, argument instantiation, and memory-aware failure recovery. Our results demonstrate that CUA-Skill substantially improves execution success rates and robustness on challenging end-to-end agent benchmarks, establishing a strong foundation for future computer-using agent development. On WindowsAgentArena, CUA-Skill Agent achieves state-of-the-art 57.5% (best of three) successful rate while being significantly more efficient than prior and concurrent approaches. The project page is available at https://microsoft.github.io/cua_skill/.

---

### An AI Framework for Microanastomosis Motion Assessment

Authors: Yan Meng, Eduardo J. Torres-Rodríguez, Marcelle Altshuler, Nishanth Gowda, Arhum Naeem, Recai Yilmaz, Omar Arnaout, Daniel A. Donoho
Published: 2026-01-28
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21120v1

Proficiency in microanastomosis is a fundamental competency across multiple microsurgical disciplines. These procedures demand exceptional precision and refined technical skills, making effective, standardized assessment methods essential. Traditionally, the evaluation of microsurgical techniques has relied heavily on the subjective judgment of expert raters. They are inherently constrained by limitations such as inter-rater variability, lack of standardized evaluation criteria, susceptibility to cognitive bias, and the time-intensive nature of manual review. These shortcomings underscore the urgent need for an objective, reliable, and automated system capable of assessing microsurgical performance with consistency and scalability. To bridge this gap, we propose a novel AI framework for the automated assessment of microanastomosis instrument handling skills. The system integrates four core components: (1) an instrument detection module based on the You Only Look Once (YOLO) architecture; (2) an instrument tracking module developed from Deep Simple Online and Realtime Tracking (DeepSORT); (3) an instrument tip localization module employing shape descriptors; and (4) a supervised classification module trained on expert-labeled data to evaluate instrument handling proficiency. Experimental results demonstrate the effectiveness of the framework, achieving an instrument detection precision of 97%, with a mean Average Precision (mAP) of 96%, measured by Intersection over Union (IoU) thresholds ranging from 50% to 95% (mAP50-95).

---

### AI-Assisted Engineering Should Track the Epistemic Status and Temporal Validity of Architectural Decisions

Authors: Sankalp Gilda, Shlok Gilda
Published: 2026-01-28
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2601.21116v1

This position paper argues that AI-assisted software engineering requires explicit mechanisms for tracking the epistemic status and temporal validity of architectural decisions. LLM coding assistants generate decisions faster than teams can validate them, yet no widely-adopted framework distinguishes conjecture from verified knowledge, prevents trust inflation through conservative aggregation, or detects when evidence expires. We propose three requirements for responsible AI-assisted engineering: (1) epistemic layers that separate unverified hypotheses from empirically validated claims, (2) conservative assurance aggregation grounded in the Gödel t-norm that prevents weak evidence from inflating confidence, and (3) automated evidence decay tracking that surfaces stale assumptions before they cause failures. We formalize these requirements as the First Principles Framework (FPF), ground its aggregation semantics in fuzzy logic, and define a quintet of invariants that any valid aggregation operator must satisfy. Our retrospective audit applying FPF criteria to two internal projects found that 20-25% of architectural decisions had stale evidence within two months, validating the need for temporal accountability. We outline research directions including learnable aggregation operators, federated evidence sharing, and SMT-based claim validation.

---

### Multi-task Code LLMs: Data Mix or Model Merge?

Authors: Mingzhi Zhu, Boris Sobolev, Rahul Krishna, Raju Pavuluri, Stacy Patterson, Michele Merler
Published: 2026-01-28
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2601.21115v1

Recent research advocates deploying smaller, specialized code LLMs in agentic frameworks alongside frontier models, sparking interest in efficient strategies for multi-task learning that balance performance, constraints, and costs. We compare two approaches for creating small, multi-task code LLMs: data mixing versus model merging. We conduct extensive experiments across two model families (Qwen Coder and DeepSeek Coder) at two scales (2B and 7B parameters), fine-tuning them for code generation and code summarization tasks. Our evaluation on HumanEval, MBPP, and CodeXGlue benchmarks reveals that model merging achieves the best overall performance at larger scale across model families, retaining 96% of specialized model performance on code generation tasks while maintaining summarization capabilities. Notably, merged models can even surpass individually fine-tuned models, with our best configuration of Qwen Coder 2.5 7B model achieving 92.7% Pass@1 on HumanEval compared to 90.9% for its task-specific fine-tuned equivalent. At a smaller scale we find instead data mixing to be a preferred strategy. We further introduce a weight analysis technique to understand how different tasks affect model parameters and their implications for merging strategies. The results suggest that careful merging and mixing strategies can effectively combine task-specific capabilities without significant performance degradation, making them ideal for resource-constrained deployment scenarios.

---

### Planner-Auditor Twin: Agentic Discharge Planning with FHIR-Based LLM Planning, Guideline Recall, Optional Caching and Self-Improvement

Authors: Kaiyuan Wu, Aditya Nagori, Rishikesan Kamaleswaran
Published: 2026-01-28
Categories: cs.AI, cs.MA
arXiv: https://arxiv.org/abs/2601.21113v1

Objective: Large language models (LLMs) show promise for clinical discharge planning, but their use is constrained by hallucination, omissions, and miscalibrated confidence. We introduce a self-improving, cache-optional Planner-Auditor framework that improves safety and reliability by decoupling generation from deterministic validation and targeted replay. Materials and Methods: We implemented an agentic, retrospective, FHIR-native evaluation pipeline using MIMIC-IV-on-FHIR. For each patient, the Planner (LLM) generates a structured discharge action plan with an explicit confidence estimate. The Auditor is a deterministic module that evaluates multi-task coverage, tracks calibration (Brier score, ECE proxies), and monitors action-distribution drift. The framework supports two-tier self-improvement: (i) within-episode regeneration when enabled, and (ii) cross-episode discrepancy buffering with replay for high-confidence, low-coverage cases. Results: While context caching improved performance over baseline, the self-improvement loop was the primary driver of gains, increasing task coverage from 32% to 86%. Calibration improved substantially, with reduced Brier/ECE and fewer high-confidence misses. Discrepancy buffering further corrected persistent high-confidence omissions during replay. Discussion: Feedback-driven regeneration and targeted replay act as effective control mechanisms to reduce omissions and improve confidence reliability in structured clinical planning. Separating an LLM Planner from a rule-based, observational Auditor enables systematic reliability measurement and safer iteration without model retraining. Conclusion: The Planner-Auditor framework offers a practical pathway toward safer automated discharge planning using interoperable FHIR data access and deterministic auditing, supported by reproducible ablations and reliability-focused evaluation.

---

### How does information access affect LLM monitors' ability to detect sabotage?

Authors: Rauno Arike, Raja Mehta Moreno, Rohan Subramani, Shubhorup Biswas, Francis Rhys Ward
Published: 2026-01-28
Categories: cs.AI, cs.SE
arXiv: https://arxiv.org/abs/2601.21112v2

Frontier language model agents can exhibit misaligned behaviors, including deception, exploiting reward hacks, and pursuing hidden objectives. To control potentially misaligned agents, we can use LLMs themselves to monitor for misbehavior. In this paper, we study how information access affects LLM monitor performance. While one might expect that monitors perform better when they have access to more of the monitored agents' reasoning and actions, we find that contemporary systems often perform better with less information, a phenomenon we call the less-is-more effect for automated oversight. We demonstrate this phenomenon, and analyze the conditions under which it occurs, in three evaluation environments where agents must conduct sabotage while evading monitors. Leveraging this effect, we introduce extract-and-evaluate (EaE) monitoring -- a new hierarchical approach where one monitor isolates relevant excerpts from the monitored agent's trajectory and another monitor scores these excerpts. In BigCodeBench-Sabotage with a GPT-4.1-mini monitor, EaE improves sabotage detection rates by 16.8 percentage points over the next-best approach. In other settings, EaE either outperforms or is competitive with baselines. In addition, we find that agents unaware of being monitored can be caught much more easily and that monitors scoring longer reasoning traces gain more from information filtering. Lastly, we conduct a cost-performance analysis and find that Gemini and Claude models with monitoring techniques that involve information filtering occupy much of the Pareto frontier.

---

### ChunkWise LoRA: Adaptive Sequence Partitioning for Memory-Efficient Low-Rank Adaptation and Accelerated LLM Inference

Authors: Ketan Thakkar, Maitreyi Chatterjee, Ramasubramanian Balasubramanian, Achyuthan Jootoo, Rajendra Ugrani
Published: 2026-01-28
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.21109v1

Recent advances in low-rank adaptation (LoRA) have enabled efficient fine-tuning of large language models (LLMs) with minimal additional parameters. However, existing LoRA methods apply static rank configurations uniformly across all input tokens, ignoring variation in token complexity and computational requirements. In this work, we propose ChunkWise LoRA, a dynamic and adaptive approach that partitions sequences into variable-length chunks based on token complexity and assigns each chunk a tailored low-rank configuration. Our system introduces a runtime scheduler that estimates token difficulty, performs adaptive chunking, and selects per-chunk LoRA rank and scaling using a rank-ladder mechanism. To preserve output consistency, we further introduce a boundary-safe composition module and integrate policy-driven KV-cache strategies. Experiments on benchmark datasets such as Wikitext-103 and SQuAD demonstrate that ChunkWise LoRA achieves up to 34\% lower latency and 38% memory reduction compared to baseline LoRA, while maintaining or improving task performance metrics like BLEU, EM, and perplexity. The proposed framework remains fully compatible with existing transformer architectures and inference frameworks, providing a practical solution for real-world deployment of parameter-efficient LLMs.

---

### SteerEval: A Framework for Evaluating Steerability with Natural Language Profiles for Recommendation

Authors: Joyce Zhou, Weijie Zhou, Doug Turnbull, Thorsten Joachims
Published: 2026-01-28
Categories: cs.IR, cs.AI, cs.CL, cs.HC
arXiv: https://arxiv.org/abs/2601.21105v1

Natural-language user profiles have recently attracted attention not only for improved interpretability, but also for their potential to make recommender systems more steerable. By enabling direct editing, natural-language profiles allow users to explicitly articulate preferences that may be difficult to infer from past behavior. However, it remains unclear whether current natural-language-based recommendation methods can follow such steering commands. While existing steerability evaluations have shown some success for well-recognized item attributes (e.g., movie genres), we argue that these benchmarks fail to capture the richer forms of user control that motivate steerable recommendations. To address this gap, we introduce SteerEval, an evaluation framework designed to measure more nuanced and diverse forms of steerability by using interventions that range from genres to content-warning for movies. We assess the steerability of a family of pretrained natural-language recommenders, examine the potential and limitations of steering on relatively niche topics, and compare how different profile and recommendation interventions impact steering effectiveness. Finally, we offer practical design suggestions informed by our findings and discuss future steps in steerable recommender design.

---

### Multilevel and Sequential Monte Carlo for Training-Free Diffusion Guidance

Authors: Aidan Gleich, Scott C. Schmidler
Published: 2026-01-28
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2601.21104v1

We address the problem of accurate, training-free guidance for conditional generation in trained diffusion models. Existing methods typically rely on point-estimates to approximate the posterior score, often resulting in biased approximations that fail to capture multimodality inherent to the reverse process of diffusion models. We propose a sequential Monte Carlo (SMC) framework that constructs an unbiased estimator of $p_θ(y|x_t)$ by integrating over the full denoising distribution via Monte Carlo approximation. To ensure computational tractability, we incorporate variance-reduction schemes based on Multi-Level Monte Carlo (MLMC). Our approach achieves new state-of-the-art results for training-free guidance on CIFAR-10 class-conditional generation, achieving $95.6\%$ accuracy with $3\times$ lower cost-per-success than baselines. On ImageNet, our algorithm achieves $1.5\times$ cost-per-success advantage over existing methods.

---

### Magellan: Autonomous Discovery of Novel Compiler Optimization Heuristics with AlphaEvolve

Authors: Hongzheng Chen, Alexander Novikov, Ngân Vũ, Hanna Alam, Zhiru Zhang, Aiden Grossman, Mircea Trofin, Amir Yazdanbakhsh
Published: 2026-01-28
Categories: cs.AI, cs.LG, cs.PL
arXiv: https://arxiv.org/abs/2601.21096v1

Modern compilers rely on hand-crafted heuristics to guide optimization passes. These human-designed rules often struggle to adapt to the complexity of modern software and hardware and lead to high maintenance burden. To address this challenge, we present Magellan, an agentic framework that evolves the compiler pass itself by synthesizing executable C++ decision logic. Magellan couples an LLM coding agent with evolutionary search and autotuning in a closed loop of generation, evaluation on user-provided macro-benchmarks, and refinement, producing compact heuristics that integrate directly into existing compilers. Across several production optimization tasks, Magellan discovers policies that match or surpass expert baselines. In LLVM function inlining, Magellan synthesizes new heuristics that outperform decades of manual engineering for both binary-size reduction and end-to-end performance. In register allocation, it learns a concise priority rule for live-range processing that matches intricate human-designed policies on a large-scale workload. We also report preliminary results on XLA problems, demonstrating portability beyond LLVM with reduced engineering effort.

---

### Responsible AI: The Good, The Bad, The AI

Authors: Akbar Anbar Jafari, Cagri Ozcinar, Gholamreza Anbarjafari
Published: 2026-01-28
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21095v1

The rapid proliferation of artificial intelligence across organizational contexts has generated profound strategic opportunities while introducing significant ethical and operational risks. Despite growing scholarly attention to responsible AI, extant literature remains fragmented and is often adopting either an optimistic stance emphasizing value creation or an excessively cautious perspective fixated on potential harms. This paper addresses this gap by presenting a comprehensive examination of AI's dual nature through the lens of strategic information systems. Drawing upon a systematic synthesis of the responsible AI literature and grounded in paradox theory, we develop the Paradox-based Responsible AI Governance (PRAIG) framework that articulates: (1) the strategic benefits of AI adoption, (2) the inherent risks and unintended consequences, and (3) governance mechanisms that enable organizations to navigate these tensions. Our framework advances theoretical understanding by conceptualizing responsible AI governance as the dynamic management of paradoxical tensions between value creation and risk mitigation. We provide formal propositions demonstrating that trade-off approaches amplify rather than resolve these tensions, and we develop a taxonomy of paradox management strategies with specified contingency conditions. For practitioners, we offer actionable guidance for developing governance structures that neither stifle innovation nor expose organizations to unacceptable risks. The paper concludes with a research agenda for advancing responsible AI governance scholarship.

---

### Safety Generalization Under Distribution Shift in Safe Reinforcement Learning: A Diabetes Testbed

Authors: Minjae Kwon, Josephine Lamp, Lu Feng
Published: 2026-01-28
Categories: cs.LG, cs.AI, eess.SY
arXiv: https://arxiv.org/abs/2601.21094v2

Safe Reinforcement Learning (RL) algorithms are typically evaluated under fixed training conditions. We investigate whether training-time safety guarantees transfer to deployment under distribution shift, using diabetes management as a safety-critical testbed. We benchmark safe RL algorithms on a unified clinical simulator and reveal a safety generalization gap: policies satisfying constraints during training frequently violate safety requirements on unseen patients. We demonstrate that test-time shielding, which filters unsafe actions using learned dynamics models, effectively restores safety across algorithms and patient populations. Across eight safe RL algorithms, three diabetes types, and three age groups, shielding achieves Time-in-Range gains of 13--14\% for strong baselines such as PPO-Lag and CPO while reducing clinical risk index and glucose variability. Our simulator and benchmark provide a platform for studying safety under distribution shift in safety-critical control domains. Code is available at https://github.com/safe-autonomy-lab/GlucoSim and https://github.com/safe-autonomy-lab/GlucoAlg.

---

### High-dimensional learning dynamics of multi-pass Stochastic Gradient Descent in multi-index models

Authors: Zhou Fan, Leda Wang
Published: 2026-01-28
Categories: stat.ML, cs.LG, math.OC, math.PR, math.ST
arXiv: https://arxiv.org/abs/2601.21093v2

We study the learning dynamics of a multi-pass, mini-batch Stochastic Gradient Descent (SGD) procedure for empirical risk minimization in high-dimensional multi-index models with isotropic random data. In an asymptotic regime where the sample size $n$ and data dimension $d$ increase proportionally, for any sub-linear batch size $κ\asymp n^α$ where $α\in (0,1)$, and for a commensurate ``critical'' scaling of the learning rate, we provide an asymptotically exact characterization of the coordinate-wise dynamics of SGD. This characterization takes the form of a system of dynamical mean-field equations, driven by a scalar Poisson jump process that represents the asymptotic limit of SGD sampling noise. We develop an analogous characterization of the Stochastic Modified Equation (SME) which provides a Gaussian diffusion approximation to SGD. Our analyses imply that the limiting dynamics for SGD are the same for any batch size scaling $α\in (0,1)$, and that under a commensurate scaling of the learning rate, dynamics of SGD, SME, and gradient flow are mutually distinct, with those of SGD and SME coinciding in the special case of a linear model. We recover a known dynamical mean-field characterization of gradient flow in a limit of small learning rate, and of one-pass/online SGD in a limit of increasing sample size $n/d \to \infty$.

---

### MapPFN: Learning Causal Perturbation Maps in Context

Authors: Marvin Sextro, Weronika Kłos, Gabriel Dernbach
Published: 2026-01-28
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21092v3

Planning effective interventions in biological systems requires treatment-effect models that adapt to unseen biological contexts by identifying their specific underlying mechanisms. Yet single-cell perturbation datasets span only a handful of biological contexts, and existing methods cannot leverage new interventional evidence at inference time to adapt beyond their training data. To meta-learn a perturbation effect estimator, we present MapPFN, a prior-data fitted network (PFN) pre-trained on a synthetic biological prior with causal interventions, decoupling pre-training from limited wet-lab data. Unlike existing methods, MapPFN uses in-context learning to map a sequence of experiments to a post-perturbation distribution, enabling a single pre-trained model to adapt to new datasets and arbitrary gene sets at inference time. Zero-shot, MapPFN identifies differentially expressed genes on par with models trained on real single-cell data, and fine-tuning further improves predictions across biological contexts. Our code, model and data are available at https://marvinsxtr.github.io/MapPFN.

---

### Deep Reinforcement Learning for Fault-Adaptive Routing in Eisenstein-Jacobi Interconnection Topologies

Authors: Mohammad Walid Charrwi, Zaid Hussain
Published: 2026-01-28
Categories: cs.DC, cs.AI
arXiv: https://arxiv.org/abs/2601.21090v1

The increasing density of many-core architectures necessitates interconnection networks that are both high-performance and fault-resilient. Eisenstein-Jacobi (EJ) networks, with their symmetric 6-regular topology, offer superior topological properties but challenge traditional routing heuristics under fault conditions. This paper evaluates three routing paradigms in faulty EJ environments: deterministic Greedy Adaptive Routing, theoretically optimal Dijkstra's algorithm, and a reinforcement learning (RL)-based approach. Using a multi-objective reward function to penalize fault proximity and reward path efficiency, the RL agent learns to navigate around clustered failures that typically induce dead-ends in greedy geometric routing. Dijkstra's algorithm establishes the theoretical performance ceiling by computing globally optimal paths with complete topology knowledge, revealing the true connectivity limits of faulty networks. Quantitative analysis at nine faulty nodes shows greedy routing catastrophically degrades to 10% effective reachability and packet delivery, while Dijkstra proves 52-54% represents the topological optimum. The RL agent achieves 94% effective reachability and 91% packet delivery, making it suitable for distributed deployment. Furthermore, throughput evaluations demonstrate that RL sustains over 90% normalized throughput across all loads, actually outperforming Dijkstra under congestion through implicit load balancing strategies. These results establish RL-based adaptive policies as a practical solution that bridges the gap between greedy's efficiency and Dijkstra's optimality, providing robust, self-healing communication in fault-prone interconnection networks without requiring the global topology knowledge or computational overhead of optimal algorithms.

---

### An efficient, accurate, and interpretable machine learning method for computing probability of failure

Authors: Jacob Zhu, Donald Estep
Published: 2026-01-28
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2601.21089v1

We introduce a novel machine learning method called the Penalized Profile Support Vector Machine based on the Gabriel edited set for the computation of the probability of failure for a complex system as determined by a threshold condition on a computer model of system behavior. The method is designed to minimize the number of evaluations of the computer model while preserving the geometry of the decision boundary that determines the probability. It employs an adaptive sampling strategy designed to strategically allocate points near the boundary determining failure and builds a locally linear surrogate boundary that remains consistent with its geometry by strategic clustering of training points. We prove two convergence results and we compare the performance of the method against a number of state of the art classification methods on four test problems. We also apply the method to determine the probability of survival using the Lotka--Volterra model for competing species.

---

### Position-invariant Fine-tuning of Speech Enhancement Models with Self-supervised Speech Representations

Authors: Amit Meghanani, Thomas Hain
Published: 2026-01-28
Categories: cs.CL, cs.SD, eess.AS
arXiv: https://arxiv.org/abs/2601.21084v1

Integrating front-end speech enhancement (SE) models with self-supervised learning (SSL)-based speech models is effective for downstream tasks in noisy conditions. SE models are commonly fine-tuned using SSL representations with mean squared error (MSE) loss between enhanced and clean speech. However, MSE is prone to exploiting positional embeddings in SSL models, allowing the objective to be minimised through positional correlations instead of content-related information. This work frames the problem as a general limitation of self-supervised representation fine-tuning and investigates it through representation-guided SE. Two strategies are considered: (1) zero-padding, previously explored in SSL pre-training but here examined in the fine-tuning setting, and (2) speed perturbations with a soft-DTW loss. Experiments show that the soft-DTW-based approach achieves faster convergence and improved downstream performance, underscoring the importance of position-invariant fine-tuning in SSL-based speech modelling.

---

### OpenSec: Measuring Incident Response Agent Calibration Under Adversarial Evidence

Authors: Jarrod Barnes
Published: 2026-01-28
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21083v3

As large language models (LLMs) improve, so do their offensive applications: frontier agents now generate working exploits for under $50 in compute (Heelan, 2026). Defensive incident response (IR) agents must keep pace, but existing benchmarks conflate action execution with correct execution, hiding calibration failures when agents process adversarial evidence. We introduce OpenSec, a dual-control reinforcement learning (RL) environment that evaluates IR agents under realistic prompt injection scenarios with execution-based scoring: time-to-first-containment (TTFC), evidence-gated action rate (EGAR), blast radius, and per-tier injection violation rates. Evaluating four frontier models on 40 standard-tier episodes each, we find consistent over-triggering: GPT-5.2 executes containment in 100% of episodes with 82.5% false positive rate, acting at step 4 before gathering sufficient evidence. Claude Sonnet 4.5 shows partial calibration (62.5% containment, 45% FP, TTFC of 10.6), suggesting calibration is not reliably present across frontier models. All models correctly identify the ground-truth threat when they act; the calibration gap is not in detection but in restraint. Code available at https://github.com/jbarnes850/opensec-env.

---

### LOCUS: Low-Dimensional Model Embeddings for Efficient Model Exploration, Comparison, and Selection

Authors: Shivam Patel, William Cocke, Gauri Joshi
Published: 2026-01-28
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21082v1

The rapidly growing ecosystem of Large Language Models (LLMs) makes it increasingly challenging to manage and utilize the vast and dynamic pool of models effectively. We propose LOCUS, a method that produces low-dimensional vector embeddings that compactly represent a language model's capabilities across queries. LOCUS is an attention-based approach that generates embeddings by a deterministic forward pass over query encodings and evaluation scores via an encoder model, enabling seamless incorporation of new models to the pool and refinement of existing model embeddings without having to perform any retraining. We additionally train a correctness predictor that uses model embeddings and query encodings to achieve state-of-the-art routing accuracy on unseen queries. Experiments show that LOCUS needs up to 4.8x fewer query evaluation samples than baselines to produce informative and robust embeddings. Moreover, the learned embedding space is geometrically meaningful: proximity reflects model similarity, enabling a range of downstream applications including model comparison and clustering, model portfolio selection, and resilient proxies of unavailable models.

---

### Shape of Thought: Progressive Object Assembly via Visual Chain-of-Thought

Authors: Yu Huo, Siyu Zhang, Kun Zeng, Haoyue Liu, Owen Lee, Junlin Chen, Yuquan Lu, Yifu Guo, Yaodong Liang, Xiaoying Tang
Published: 2026-01-28
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21081v2

Multimodal models for text-to-image generation have achieved strong visual fidelity, yet they remain brittle under compositional structural constraints, notably generative numeracy, attribute binding, and part-level relations. To address these challenges, we propose Shape-of-Thought (SoT), a visual CoT framework for process-supervised progressive shape assembly in the rendered 2D domain, without external engines at inference time. SoT trains a unified multimodal autoregressive model to generate interleaved textual plans and rendered intermediate states, helping the model capture shape-assembly logic without producing explicit geometric representations. Unlike text-only CoT, each decision is grounded in a rendered state, making counts, attachments, topology, and intermediate part-addition errors inspectable across the trajectory. To support this paradigm, we introduce SoT-26K, a large-scale dataset of grounded assembly traces derived from part-based CAD hierarchies, and T2S-CompBench, a benchmark for evaluating structural integrity and trace faithfulness. Fine-tuning on SoT-26K achieves 88.4% on component numeracy and 84.8% on structural topology, outperforming direct generation by +24.2 points on component numeracy and +19.3 points on structural topology. SoT establishes a transparent testbed for rendered-domain structure-aware generation. The code is available at https://github.com/yuhuo03/Shape-of-Thought.

---

### Parametric Hyperbolic Conservation Laws: A Unified Framework for Conservation, Entropy Stability, and Hyperbolicity

Authors: Lizuo Liu, Lu Zhang, Anne Gelb
Published: 2026-01-28
Categories: math.NA, cs.LG, math-ph
arXiv: https://arxiv.org/abs/2601.21080v1

We propose a parametric hyperbolic conservation law (SymCLaw) for learning hyperbolic systems directly from data while ensuring conservation, entropy stability, and hyperbolicity by design. Unlike existing approaches that typically enforce only conservation or rely on prior knowledge of the governing equations, our method parameterizes the flux functions in a form that guarantees real eigenvalues and complete eigenvectors of the flux Jacobian, thereby preserving hyperbolicity. At the same time, we embed entropy-stable design principles by jointly learning a convex entropy function and its associated flux potential, ensuring entropy dissipation and the selection of physically admissible weak solutions. A corresponding entropy-stable numerical flux scheme provides compatibility with standard discretizations, allowing seamless integration into classical solvers. Numerical experiments on benchmark problems, including Burgers, shallow water, Euler, and KPP equations, demonstrate that SymCLaw generalizes to unseen initial conditions, maintains stability under noisy training data, and achieves accurate long-time predictions, highlighting its potential as a principled foundation for data-driven modeling of hyperbolic conservation laws.

---

### Towards Mitigating Modality Bias in Vision-Language Models for Temporal Action Localization

Authors: Jiaqi Li, Guangming Wang, Shuntian Zheng, Minzhe Ni, Xiaoman Lu, Guanghui Ye, Yu Guan
Published: 2026-01-28
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21078v3

Temporal Action Localization (TAL) requires identifying both the boundaries and categories of actions in untrimmed videos. While vision-language models (VLMs) offer rich semantics to complement visual evidence, existing approaches tend to overemphasize linguistic priors at the expense of visual performance, leading to a pronounced modality bias. We propose ActionVLM, a vision-language aggregation framework that systematically mitigates modality bias in TAL. Our key insight is to preserve vision as the dominant signal while adaptively exploiting language only when beneficial. To this end, we introduce (i) a debiasing reweighting module that estimates the language advantage-the incremental benefit of language over vision-only predictions-and dynamically reweights language modality accordingly, and (ii) a residual aggregation strategy that treats language as a complementary refinement rather than the primary driver. This combination alleviates modality bias, reduces overconfidence from linguistic priors, and strengthens temporal reasoning. Experiments on THUMOS14 show that our model outperforms state-of-the-art by up to 3.2% mAP. Our code is available at https://github.com/JiaqiLi404/ActionVLM

---

### Multi-modal Imputation for Alzheimer's Disease Classification

Authors: Abhijith Shaji, Tamoghna Chattopadhyay, Sophia I. Thomopoulos, Greg Ver Steeg, Paul M. Thompson, Jose-Luis Ambite
Published: 2026-01-28
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21076v1

Deep learning has been successful in predicting neurodegenerative disorders, such as Alzheimer's disease, from magnetic resonance imaging (MRI). Combining multiple imaging modalities, such as T1-weighted (T1) and diffusion-weighted imaging (DWI) scans, can increase diagnostic performance. However, complete multimodal datasets are not always available. We use a conditional denoising diffusion probabilistic model to impute missing DWI scans from T1 scans. We perform extensive experiments to evaluate whether such imputation improves the accuracy of uni-modal and bi-modal deep learning models for 3-way Alzheimer's disease classification-cognitively normal, mild cognitive impairment, and Alzheimer's disease. We observe improvements in several metrics, particularly those sensitive to minority classes, for several imputation configurations.

---

### Towards Comprehensive Benchmarking Infrastructure for LLMs In Software Engineering

Authors: Daniel Rodriguez-Cardenas, Xiaochang Li, Marcos Macedo, Antonio Mastropaolo, Dipin Khati, Yuan Tian, Huajie Shao, Denys Poshyvanyk
Published: 2026-01-28
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2601.21070v1

Large language models for code are advancing fast, yet our ability to evaluate them lags behind. Current benchmarks focus on narrow tasks and single metrics, which hide critical gaps in robustness, interpretability, fairness, efficiency, and real-world usability. They also suffer from inconsistent data engineering practices, limited software engineering context, and widespread contamination issues. To understand these problems and chart a path forward, we combined an in-depth survey of existing benchmarks with insights gathered from a dedicated community workshop. We identified three core barriers to reliable evaluation: the absence of software-engineering-rich datasets, overreliance on ML-centric metrics, and the lack of standardized, reproducible data pipelines. Building on these findings, we introduce BEHELM, a holistic benchmarking infrastructure that unifies software-scenario specification with multi-metric evaluation. BEHELM provides a structured way to assess models across tasks, languages, input and output granularities, and key quality dimensions. Our goal is to reduce the overhead currently required to construct benchmarks while enabling a fair, realistic, and future-proof assessment of LLMs in software engineering.

---

### Out-of-Distribution Generalization in Graph Foundation Models

Authors: Haoyang Li, Haibo Chen, Xin Wang, Wenwu Zhu
Published: 2026-01-28
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21067v1

Graphs are a fundamental data structure for representing relational information in domains such as social networks, molecular systems, and knowledge graphs. However, graph learning models often suffer from limited generalization when applied beyond their training distributions. In practice, distribution shifts may arise from changes in graph structure, domain semantics, available modalities, or task formulations. To address these challenges, graph foundation models (GFMs) have recently emerged, aiming to learn general-purpose representations through large-scale pretraining across diverse graphs and tasks. In this survey, we review recent progress on GFMs from the perspective of out-of-distribution (OOD) generalization. We first discuss the main challenges posed by distribution shifts in graph learning and outline a unified problem setting. We then organize existing approaches based on whether they are designed to operate under a fixed task specification or to support generalization across heterogeneous task formulations, and summarize the corresponding OOD handling strategies and pretraining objectives. Finally, we review common evaluation protocols and discuss open directions for future research. To the best of our knowledge, this paper is the first survey for OOD generalization in GFMs.

---

### BadDet+: Robust Backdoor Attacks for Object Detection

Authors: Kealan Dunnett, Reza Arablouei, Dimity Miller, Volkan Dedeoglu, Raja Jurdak
Published: 2026-01-28
Categories: cs.CV, cs.CR
arXiv: https://arxiv.org/abs/2601.21066v1

Backdoor attacks pose a severe threat to deep learning, yet their impact on object detection remains poorly understood compared to image classification. While attacks have been proposed, we identify critical weaknesses in existing detection-based methods, specifically their reliance on unrealistic assumptions and a lack of physical validation. To bridge this gap, we introduce BadDet+, a penalty-based framework that unifies Region Misclassification Attacks (RMA) and Object Disappearance Attacks (ODA). The core mechanism utilizes a log-barrier penalty to suppress true-class predictions for triggered inputs, resulting in (i) position and scale invariance, and (ii) enhanced physical robustness. On real-world benchmarks, BadDet+ achieves superior synthetic-to-physical transfer compared to existing RMA and ODA baselines while preserving clean performance. Theoretical analysis confirms the proposed penalty acts within a trigger-specific feature subspace, reliably inducing attacks without degrading standard inference. These results highlight significant vulnerabilities in object detection and the necessity for specialized defenses.

---

### Textual Equilibrium Propagation for Deep Compound AI Systems

Authors: Minghui Chen, Wenlong Deng, James Zou, Han Yu, Xiaoxiao Li
Published: 2026-01-28
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21064v3

Large language models (LLMs) are increasingly deployed as part of compound AI systems that coordinate multiple modules (e.g., retrievers, tools, verifiers) over long-horizon workflows. Recent approaches that propagate textual feedback globally (e.g., TextGrad) make it feasible to optimize such pipelines, but we find that performance degrades as system depth grows. In particular, long-horizon agentic workflows exhibit two depth-scaling failure modes: 1) exploding textual gradient, where textual feedback grows exponentially with depth, leading to prohibitively long message and amplifies evaluation biases; and 2) vanishing textual gradient, where limited long-context ability causes models overemphasize partial feedback and compression of lengthy feedback causes downstream messages to lose specificity gradually as they propagate many hops upstream. To mitigate these issues, we introduce Textual Equilibrium Propagation (TEP), a local learning principle inspired by Equilibrium Propagation in energy-based models. TEP includes two phases: 1) a free phase where a local LLM critics iteratively refine prompts until reaching equilibrium (no further improvements are suggested); and 2) a nudged phase which applies proximal prompt edits with bounded modification intensity, using task-level objectives that propagate via forward signaling rather than backward feedback chains. This design supports local prompt optimization followed by controlled adaptation toward global goals without the computational burden and signal degradation of global textual backpropagation. Across long-horizon QA benchmarks and multi-agent tool-use dataset, TEP consistently improves accuracy and efficiency over global propagation methods such as TextGrad. The gains grows with depth, while preserving the practicality of black-box LLM components in deep compound AI system.

---