# Research Papers — 2026-01 · bundle page 108 of 189

> Full-text markdown of papers #5351–5400 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/108
[← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/107.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-01/page/109.md)

### Handling Missing Modalities in Multimodal Survival Prediction for Non-Small Cell Lung Cancer

Authors: Filippo Ruffini, Camillo Maria Caruso, Claudia Tacconi, Lorenzo Nibid, Francesca Miccolis, Marta Lovino, Carlo Greco, Edy Ippolito, Michele Fiore, Alessio Cortellini, Bruno Beomonte Zobel, Giuseppe Perrone et al.
Published: 2026-01-15
Categories: cs.CV, cs.AI, cs.MM
arXiv: https://arxiv.org/abs/2601.10386v2

Accurate survival prediction in Non-Small Cell Lung Cancer (NSCLC) requires integrating clinical, radiological, and histopathological data. Multimodal Deep Learning (MDL) can improve precision prognosis, but small cohorts and missing modalities limit its clinical applicability, as conventional approaches enforce complete case filtering or imputation. We present a missing-aware multimodal survival framework that combines Computed Tomography (CT), Whole-Slide Histopathology Images (WSI), and structured clinical variables for overall survival modeling in unresectable stage II-III NSCLC. The framework uses Foundation Models (FMs) for modality-specific feature extraction and a missing-aware encoding strategy that enables intermediate multimodal fusion under naturally incomplete modality profiles. By design, the architecture processes all available data without dropping patients during training or inference. Intermediate fusion outperforms unimodal baselines and both early and late fusion strategies, with the trimodal configuration reaching a C-index of 74.42. Modality-importance analyses show that the fusion model adapts its reliance on each data stream according to representation informativeness, shaped by the alignment between FM pretraining objectives and the survival task. The learned risk scores produce clinically meaningful stratification of disease progression and metastatic risk, with statistically significant log-rank tests across all modality combinations, supporting the translational relevance of the proposed framework.

---

### Global Context Compression with Interleaved Vision-Text Transformation

Authors: Dian Jiao, Jiaxin Duan, Shuai Zhao, Jiabing Leng, Yiran Zhang, Feng Huang
Published: 2026-01-15
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2601.10378v2

Recent achievements of vision-language models in end-to-end OCR point to a new avenue for low-loss compression of textual information. This motivates earlier works that render the Transformer's input into images for prefilling, which effectively reduces the number of tokens through visual encoding, thereby alleviating the quadratically increased Attention computations. However, this partial compression fails to save computational or memory costs at token-by-token inference. In this paper, we investigate global context compression, which saves tokens at both prefilling and inference stages. Consequently, we propose VIST2, a novel Transformer that interleaves input text chunks alongside their visual encoding, while depending exclusively on visual tokens in the pre-context to predict the next text token distribution. Around this idea, we render text chunks into sketch images and train VIST2 in multiple stages, starting from curriculum-scheduled pretraining for optical language modeling, followed by modal-interleaved instruction tuning. We conduct extensive experiments using VIST2 families scaled from 0.6B to 8B to explore the training recipe and hyperparameters. With a 4$\times$ compression ratio, the resulting models demonstrate significant superiority over baselines on long writing tasks, achieving, on average, a 3$\times$ speedup in first-token generation, 77% reduction in memory usage, and 74% reduction in FLOPS. Our codes and datasets will be public to support further studies.

---

### Towards Efficient Low-rate Image Compression with Frequency-aware Diffusion Prior Refinement

Authors: Yichong Xia, Yimin Zhou, Jinpeng Wang, Bin Chen
Published: 2026-01-15
Categories: cs.CV, cs.AI, cs.MM
arXiv: https://arxiv.org/abs/2601.10373v1

Recent advancements in diffusion-based generative priors have enabled visually plausible image compression at extremely low bit rates. However, existing approaches suffer from slow sampling processes and suboptimal bit allocation due to fragmented training paradigms. In this work, we propose Accelerate \textbf{Diff}usion-based Image Compression via \textbf{C}onsistency Prior \textbf{R}efinement (DiffCR), a novel compression framework for efficient and high-fidelity image reconstruction. At the heart of DiffCR is a Frequency-aware Skip Estimation (FaSE) module that refines the $ε$-prediction prior from a pre-trained latent diffusion model and aligns it with compressed latents at different timesteps via Frequency Decoupling Attention (FDA). Furthermore, a lightweight consistency estimator enables fast \textbf{two-step decoding} by preserving the semantic trajectory of diffusion sampling. Without updating the backbone diffusion model, DiffCR achieves substantial bitrate savings (27.2\% BD-rate (LPIPS) and 65.1\% BD-rate (PSNR)) and over $10\times$ speed-up compared to SOTA diffusion-based compression baselines.

---

### Fine-Grained Human Pose Editing Assessment via Layer-Selective MLLMs

Authors: Ningyu Sun, Zhaolin Cai, Zitong Xu, Peihang Chen, Huiyu Duan, Yichao Yan, Xiongkuo Min, Xiaokang Yang
Published: 2026-01-15
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.10369v2

Text-guided human pose editing has gained significant traction in AIGC applications. However,it remains plagued by structural anomalies and generative artifacts. Existing evaluation metrics often isolate authenticity detection from quality assessment, failing to provide fine-grained insights into pose-specific inconsistencies. To address these limitations, we introduce HPE-Bench, a specialized benchmark comprising 1,700 standardized samples from 17 state-of-the-art editing models, offering both authenticity labels and multi-dimensional quality scores. Furthermore, we propose a unified framework based on layer-selective multimodal large language models (MLLMs). By employing contrastive LoRA tuning and a novel layer sensitivity analysis (LSA) mechanism, we identify the optimal feature layer for pose evaluation. Our framework achieves superior performance in both authenticity detection and multi-dimensional quality regression, effectively bridging the gap between forensic detection and quality assessment.

---

### PLGC: Pseudo-Labeled Graph Condensation

Authors: Jay Nandy, Arnab Kumar Mondal, Anuj Rathore, Mahesh Chandran
Published: 2026-01-15
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.10358v1

Large graph datasets make training graph neural networks (GNNs) computationally costly. Graph condensation methods address this by generating small synthetic graphs that approximate the original data. However, existing approaches rely on clean, supervised labels, which limits their reliability when labels are scarce, noisy, or inconsistent. We propose Pseudo-Labeled Graph Condensation (PLGC), a self-supervised framework that constructs latent pseudo-labels from node embeddings and optimizes condensed graphs to match the original graph's structural and feature statistics -- without requiring ground-truth labels. PLGC offers three key contributions: (1) A diagnosis of why supervised condensation fails under label noise and distribution shift. (2) A label-free condensation method that jointly learns latent prototypes and node assignments. (3) Theoretical guarantees showing that pseudo-labels preserve latent structural statistics of the original graph and ensure accurate embedding alignment. Empirically, across node classification and link prediction tasks, PLGC achieves competitive performance with state-of-the-art supervised condensation methods on clean datasets and exhibits substantial robustness under label noise, often outperforming all baselines by a significant margin. Our findings highlight the practical and theoretical advantages of self-supervised graph condensation in noisy or weakly-labeled environments.

---

### EvoMorph: Counterfactual Explanations for Continuous Time-Series Extrinsic Regression Applied to Photoplethysmography

Authors: Mesut Ceylan, Alexis Tabin, Patrick Langer, Elgar Fleisch, Filipe Barata
Published: 2026-01-15
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.10356v1

Wearable devices enable continuous, population-scale monitoring of physiological signals, such as photoplethysmography (PPG), creating new opportunities for data-driven clinical assessment. Time-series extrinsic regression (TSER) models increasingly leverage PPG signals to estimate clinically relevant outcomes, including heart rate, respiratory rate, and oxygen saturation. For clinical reasoning and trust, however, single point estimates alone are insufficient: clinicians must also understand whether predictions are stable under physiologically plausible variations and to what extent realistic, attainable changes in physiological signals would meaningfully alter a model's prediction. Counterfactual explanations (CFE) address these "what-if" questions, yet existing time series CFE generation methods are largely restricted to classification, overlook waveform morphology, and often produce physiologically implausible signals, limiting their applicability to continuous biomedical time series. To address these limitations, we introduce EvoMorph, a multi-objective evolutionary framework for generating physiologically plausible and diverse CFE for TSER applications. EvoMorph optimizes morphology-aware objectives defined on interpretable signal descriptors and applies transformations to preserve the waveform structure. We evaluated EvoMorph on three PPG datasets (heart rate, respiratory rate, and oxygen saturation) against a nearest-unlike-neighbor baseline. In addition, in a case study, we evaluated EvoMorph as a tool for uncertainty quantification by relating counterfactual sensitivity to bootstrap-ensemble uncertainty and data-density measures. Overall, EvoMorph enables the generation of physiologically-aware counterfactuals for continuous biomedical signals and supports uncertainty-aware interpretability, advancing trustworthy model analysis for clinical time-series applications.

---

### Unlocking Implicit Experience: Synthesizing Tool-Use Trajectories from Text

Authors: Zhihao Xu, Rumei Li, Jiahuan Li, Rongxiang Weng, Jingang Wang, Xunliang Cai, Xiting Wang
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10355v1

Enabling Large Language Models (LLMs) to effectively utilize tools in multi-turn interactions is essential for building capable autonomous agents. However, acquiring diverse and realistic multi-turn tool-use data remains a significant challenge. In this work, we propose a novel text-based paradigm. We observe that textual corpora naturally contain rich, multi-step problem-solving experiences, which can serve as an untapped, scalable, and authentic data source for multi-turn tool-use tasks. Based on this insight, we introduce GEM, a data synthesis pipeline that enables the generation and extraction of multi-turn tool-use trajectories from text corpora through a four-stage process: relevance filtering, workflow & tool extraction, trajectory grounding, and complexity refinement. To reduce the computational cost, we further train a specialized Trajectory Synthesizer via supervised fine-tuning. This model distills the complex generation pipeline into an efficient, end-to-end trajectory generator. Experiments demonstrate that our GEM-32B achieve a 16.5% improvement on the BFCL V3 Multi-turn benchmark. Our models partially surpass the performance of models trained on τ - bench (Airline and Retail) in-domain data, highlighting the superior generalization capability derived from our text-based synthesis paradigm. Notably, our Trajectory Synthesizer matches the quality of the full pipeline while significantly reducing inference latency and costs.

---

### SuS: Strategy-aware Surprise for Intrinsic Exploration

Authors: Mark Kashirskiy, Ilya Makarov
Published: 2026-01-15
Categories: cs.LG, cs.AI, cs.CL, cs.GT
arXiv: https://arxiv.org/abs/2601.10349v1

We propose Strategy-aware Surprise (SuS), a novel intrinsic motivation framework that uses pre-post prediction mismatch as a novelty signal for exploration in reinforcement learning. Unlike traditional curiosity-driven methods that rely solely on state prediction error, SuS introduces two complementary components: Strategy Stability (SS) and Strategy Surprise (SuS). SS measures consistency in behavioral strategy across temporal steps, while SuS captures unexpected outcomes relative to the agent's current strategy representation. Our combined reward formulation leverages both signals through learned weighting coefficients. We evaluate SuS on mathematical reasoning tasks using large language models, demonstrating significant improvements in both accuracy and solution diversity. Ablation studies confirm that removing either component results in at least 10% performance degradation, validating the synergistic nature of our approach. SuS achieves 17.4% improvement in Pass@1 and 26.4% improvement in Pass@5 compared to baseline methods, while maintaining higher strategy diversity throughout training.

---

### Training-Trajectory-Aware Token Selection

Authors: Zhanming Shen, Jiaqi Hu, Zeyu Qin, Hao Chen, Wentao Ye, Zenan Huang, Yihong Zhuang, Guoshan Lu, Junlin Zhou, Junbo Zhao
Published: 2026-01-15
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2601.10348v2

Efficient distillation is a key pathway for converting expensive reasoning capability into deployable efficiency, yet in the frontier regime where the student already has strong reasoning ability, naive continual distillation often yields limited gains or even degradation. We observe a characteristic training phenomenon: even as loss decreases monotonically, all performance metrics can drop sharply at almost the same bottleneck, before gradually recovering. We further uncover a token-level mechanism: confidence bifurcates into steadily increasing Imitation-Anchor Tokens that quickly anchor optimization and other yet-to-learn tokens whose confidence is suppressed until after the bottleneck. And the characteristic that these two types of tokens cannot coexist is the root cause of the failure in continual distillation. To this end, we propose Training-Trajectory-Aware Token Selection (T3S) to reconstruct the training objective at the token level, clearing the optimization path for yet-to-learn tokens. T3S yields consistent gains in both AR and dLLM settings: with only hundreds of examples, Qwen3-8B surpasses DeepSeek-R1 on competitive reasoning benchmarks, Qwen3-32B approaches Qwen3-235B, and T3-trained LLaDA-2.0-Mini exceeds its AR baseline, achieving state-of-the-art performance among all of 16B-scale no-think models.

---

### OctoBench: Benchmarking Scaffold-Aware Instruction Following in Repository-Grounded Agentic Coding

Authors: Deming Ding, Shichun Liu, Enhui Yang, Jiahang Lin, Ziying Chen, Shihan Dou, Honglin Guo, Weiyu Cheng, Pengyu Zhao, Chengjun Xiao, Qunhong Zeng, Qi Zhang et al.
Published: 2026-01-15
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2601.10343v2

Modern coding scaffolds turn LLMs into capable software agents, but their ability to follow scaffold-specified instructions remains under-examined, especially when constraints are heterogeneous and persist across interactions. To fill this gap, we introduce OctoBench, which benchmarks scaffold-aware instruction following in repository-grounded agentic coding. OctoBench includes 34 environments and 217 tasks instantiated under three scaffold types, and is paired with 7,098 objective checklist items. To disentangle solving the task from following the rules, we provide an automated observation-and-scoring toolkit that captures full trajectories and performs fine-grained checks. Experiments on eight representative models reveal a systematic gap between task-solving and scaffold-aware compliance, underscoring the need for training and evaluation that explicitly targets heterogeneous instruction following. We release the benchmark to support reproducible benchmarking and to accelerate the development of more scaffold-aware coding agents.

---

### C-GRASP: Clinically-Grounded Reasoning for Affective Signal Processing

Authors: Cheng Lin Cheng, Ting Chuan Lin, Chai Kai Chang
Published: 2026-01-15
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.10342v1

Heart rate variability (HRV) is a pivotal noninvasive marker for autonomic monitoring; however, applying Large Language Models (LLMs) to HRV interpretation is hindered by physiological hallucinations. These include respiratory sinus arrhythmia (RSA) contamination, short-data instability in nonlinear metrics, and the neglect of individualized baselines in favor of population norms. We propose C-GRASP (Clinically-Grounded Reasoning for Affective Signal Processing), a guardrailed RAG-enhanced pipeline that decomposes HRV interpretation into eight traceable reasoning steps. Central to C-GRASP is a Z-score Priority Hierarchy that enforces the weighting of individualized baseline shifts over normative statistics. The system effectively mitigates spectral hallucinations through automated RSA-aware guardrails, preventing contamination of frequency-domain indices. Evaluated on 414 trials from the DREAMER dataset, C-GRASP integrated with high-scale reasoning models (e.g., MedGemma3-thinking) achieved superior performance in 4-class emotion classification (37.3% accuracy) and a Clinical Reasoning Consistency (CRC) score of 69.6%. Ablation studies confirm that the individualized Delta Z-score module serves as the critical logical anchor, preventing the "population bias" common in native LLMs. Ultimately, C-GRASP transitions affective computing from black-box classification to transparent, evidence-based clinical decision support, paving the way for safer AI integration in biomedical engineering.

---

### Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale

Authors: Yi Liu, Weizhe Wang, Ruitao Feng, Yao Zhang, Guangquan Xu, Gelei Deng, Yuekang Li, Leo Zhang
Published: 2026-01-15
Categories: cs.CR, cs.AI, cs.CL, cs.SE
arXiv: https://arxiv.org/abs/2601.10338v1

The rise of AI agent frameworks has introduced agent skills, modular packages containing instructions and executable code that dynamically extend agent capabilities. While this architecture enables powerful customization, skills execute with implicit trust and minimal vetting, creating a significant yet uncharacterized attack surface. We conduct the first large-scale empirical security analysis of this emerging ecosystem, collecting 42,447 skills from two major marketplaces and systematically analyzing 31,132 using SkillScan, a multi-stage detection framework integrating static analysis with LLM-based semantic classification. Our findings reveal pervasive security risks: 26.1% of skills contain at least one vulnerability, spanning 14 distinct patterns across four categories: prompt injection, data exfiltration, privilege escalation, and supply chain risks. Data exfiltration (13.3%) and privilege escalation (11.8%) are most prevalent, while 5.2% of skills exhibit high-severity patterns strongly suggesting malicious intent. We find that skills bundling executable scripts are 2.12x more likely to contain vulnerabilities than instruction-only skills (OR=2.12, p<0.001). Our contributions include: (1) a grounded vulnerability taxonomy derived from 8,126 vulnerable skills, (2) a validated detection methodology achieving 86.7% precision and 82.5% recall, and (3) an open dataset and detection toolkit to support future research. These results demonstrate an urgent need for capability-based permission systems and mandatory security vetting before this attack vector is further exploited.

---

### An analytic theory of convolutional neural network inverse problems solvers

Authors: Minh Hai Nguyen, Quoc Bao Do, Edouard Pauwels, Pierre Weiss
Published: 2026-01-15
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2601.10334v2

Supervised convolutional neural networks (CNNs) are widely used to solve imaging inverse problems, achieving state-of-the-art performance in numerous applications. However, despite their empirical success, these methods are poorly understood from a theoretical perspective and often treated as black boxes. To bridge this gap, we analyze trained neural networks through the lens of the Minimum Mean Square Error (MMSE) estimator, incorporating functional constraints that capture two fundamental inductive biases of CNNs: translation equivariance and locality via finite receptive fields. Under the empirical training distribution, we derive an analytic, interpretable, and tractable formula for this constrained variant, termed Local-Equivariant MMSE (LE-MMSE). Through extensive numerical experiments across various inverse problems (denoising, inpainting, deconvolution), datasets (FFHQ, CIFAR-10, FashionMNIST), and architectures (U-Net, ResNet, PatchMLP), we demonstrate that our theory matches the neural networks outputs (PSNR $\gtrsim25$dB). Furthermore, we provide insights into the differences between \emph{physics-aware} and \emph{physics-agnostic} estimators, the impact of high-density regions in the training (patch) distribution, and the influence of other factors (dataset size, patch size, etc).

---

### Think-Then-Generate: Reasoning-Aware Text-to-Image Diffusion with LLM Encoders

Authors: Siqi Kou, Jiachun Jin, Zetong Zhou, Ye Ma, Yugang Wang, Quan Chen, Peng Jiang, Xiao Yang, Jun Zhu, Kai Yu, Zhijie Deng
Published: 2026-01-15
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.10332v1

Recent progress in text-to-image (T2I) diffusion models (DMs) has enabled high-quality visual synthesis from diverse textual prompts. Yet, most existing T2I DMs, even those equipped with large language model (LLM)-based text encoders, remain text-pixel mappers -- they employ LLMs merely as text encoders, without leveraging their inherent reasoning capabilities to infer what should be visually depicted given the textual prompt. To move beyond such literal generation, we propose the think-then-generate (T2G) paradigm, where the LLM-based text encoder is encouraged to reason about and rewrite raw user prompts; the states of the rewritten prompts then serve as diffusion conditioning. To achieve this, we first activate the think-then-rewrite pattern of the LLM encoder with a lightweight supervised fine-tuning process. Subsequently, the LLM encoder and diffusion backbone are co-optimized to ensure faithful reasoning about the context and accurate rendering of the semantics via Dual-GRPO. In particular, the text encoder is reinforced using image-grounded rewards to infer and recall world knowledge, while the diffusion backbone is pushed to produce semantically consistent and visually coherent images. Experiments show substantial improvements in factual consistency, semantic alignment, and visual realism across reasoning-based image generation and editing benchmarks, achieving 0.79 on WISE score, nearly on par with GPT-4. Our results constitute a promising step toward next-generation unified models with reasoning, expression, and demonstration capacities.

---

### Meta Dynamic Graph for Traffic Flow Prediction

Authors: Yiqing Zou, Hanning Yuan, Qianyu Yang, Ziqiang Yuan, Shuliang Wang, Sijie Ruan
Published: 2026-01-15
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.10328v1

Traffic flow prediction is a typical spatio-temporal prediction problem and has a wide range of applications. The core challenge lies in modeling the underlying complex spatio-temporal dependencies. Various methods have been proposed, and recent studies show that the modeling of dynamics is useful to meet the core challenge. While handling spatial dependencies and temporal dependencies using separate base model structures may hinder the modeling of spatio-temporal correlations, the modeling of dynamics can bridge this gap. Incorporating spatio-temporal heterogeneity also advances the main goal, since it can extend the parameter space and allow more flexibility. Despite these advances, two limitations persist: 1) the modeling of dynamics is often limited to the dynamics of spatial topology (e.g., adjacency matrix changes), which, however, can be extended to a broader scope; 2) the modeling of heterogeneity is often separated for spatial and temporal dimensions, but this gap can also be bridged by the modeling of dynamics. To address the above limitations, we propose a novel framework for traffic prediction, called Meta Dynamic Graph (MetaDG). MetaDG leverages dynamic graph structures of node representations to explicitly model spatio-temporal dynamics. This generates both dynamic adjacency matrices and meta-parameters, extending dynamic modeling beyond topology while unifying the capture of spatio-temporal heterogeneity into a single dimension. Extensive experiments on four real-world datasets validate the effectiveness of MetaDG.

---

### SRAW-Attack: Space-Reweighted Adversarial Warping Attack for SAR Target Recognition

Authors: Yiming Zhang, Weibo Qin, Yuntian Liu, Feng Wang
Published: 2026-01-15
Categories: cs.CV, eess.IV
arXiv: https://arxiv.org/abs/2601.10324v3

Synthetic aperture radar (SAR) imagery exhibits intrinsic information sparsity due to its unique electromagnetic scattering mechanism. Despite the widespread adoption of deep neural network (DNN)-based SAR automatic target recognition (SAR-ATR) systems, they remain vulnerable to adversarial examples and tend to over-rely on background regions, leading to degraded adversarial robustness. Existing adversarial attacks for SAR-ATR often require visually perceptible distortions to achieve effective performance, thereby necessitating an attack method that balances effectiveness and stealthiness. In this paper, a novel attack method termed Space-Reweighted Adversarial Warping (SRAW) is proposed, which generates adversarial examples through optimized spatial deformation with reweighted budgets across foreground and background regions. Extensive experiments demonstrate that SRAW significantly degrades the performance of state-of-the-art SAR-ATR models and consistently outperforms existing methods in terms of imperceptibility and adversarial transferability. Code is made available at https://github.com/boremycin/SAR-ATR-TransAttack.

---

### ROMA: Real-time Omni-Multimodal Assistant with Interactive Streaming Understanding

Authors: Xueyun Tian, Wei Li, Bingbing Xu, Heng Dong, Yuanzhuo Wang, Huawei Shen
Published: 2026-01-15
Categories: cs.CV, cs.CL
arXiv: https://arxiv.org/abs/2601.10323v1

Recent Omni-multimodal Large Language Models show promise in unified audio, vision, and text modeling. However, streaming audio-video understanding remains challenging, as existing approaches suffer from disjointed capabilities: they typically exhibit incomplete modality support or lack autonomous proactive monitoring. To address this, we present ROMA, a real-time omni-multimodal assistant for unified reactive and proactive interaction. ROMA processes continuous inputs as synchronized multimodal units, aligning dense audio with discrete video frames to handle granularity mismatches. For online decision-making, we introduce a lightweight speak head that decouples response initiation from generation to ensure precise triggering without task conflict. We train ROMA with a curated streaming dataset and a two-stage curriculum that progressively optimizes for streaming format adaptation and proactive responsiveness. To standardize the fragmented evaluation landscape, we reorganize diverse benchmarks into a unified suite covering both proactive (alert, narration) and reactive (QA) settings. Extensive experiments across 12 benchmarks demonstrate ROMA achieves state-of-the-art performance on proactive tasks while competitive in reactive settings, validating its robustness in unified real-time omni-multimodal understanding.

---

### An Efficient Long-Context Ranking Architecture With Calibrated LLM Distillation: Application to Person-Job Fit

Authors: Warren Jouanneau, Emma Jouffroy, Marc Palyart
Published: 2026-01-15
Categories: cs.CL, cs.IR, cs.LG, cs.SI
arXiv: https://arxiv.org/abs/2601.10321v2

Finding the most relevant person for a job proposal in real time is challenging, especially when resumes are long, structured, and multilingual. In this paper, we propose a re-ranking model based on a new generation of late cross-attention architecture, that decomposes both resumes and project briefs to efficiently handle long-context inputs with minimal computational overhead. To mitigate historical data biases, we use a generative large language model (LLM) as a teacher, generating fine-grained, semantically grounded supervision. This signal is distilled into our student model via an enriched distillation loss function. The resulting model produces skill-fit scores that enable consistent and interpretable person-job matching. Experiments on relevance, ranking, and calibration metrics demonstrate that our approach outperforms state-of-the-art baselines.

---

### Boundary-Aware NL2SQL: Integrating Reliability through Hybrid Reward and Data Synthesis

Authors: Songsong Tian, Kongsheng Zhuo, Zhendong Wang, Rong Shen, Shengtao Zhang, Yong Wu
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10318v1

In this paper, we present BAR-SQL (Boundary-Aware Reliable NL2SQL), a unified training framework that embeds reliability and boundary awareness directly into the generation process. We introduce a Seed Mutation data synthesis paradigm that constructs a representative enterprise corpus, explicitly encompassing multi-step analytical queries alongside boundary cases including ambiguity and schema limitations. To ensure interpretability, we employ Knowledge-Grounded Reasoning Synthesis, which produces Chain-of-Thought traces explicitly anchored in schema metadata and business rules. The model is trained through a two-stage process: Supervised Fine-Tuning (SFT) followed by Reinforcement Learning via Group Relative Policy Optimization. We design a Task-Conditioned Hybrid Reward mechanism that simultaneously optimizes SQL execution accuracy-leveraging Abstract Syntax Tree analysis and dense result matching-and semantic precision in abstention responses. To evaluate reliability alongside generation accuracy, we construct and release Ent-SQL-Bench, which jointly assesse SQL precision and boundary-aware abstention across ambiguous and unanswerable queries. Experimental results on this benchmark demonstrate that BAR-SQL achieves 91.48% average accuracy, outperforming leading proprietary models, including Claude 4.5 Sonnet and GPT-5, in both SQL generation quality and boundary-aware abstention capability. The source code and benchmark are available anonymously at: https://github.com/TianSongS/BAR-SQL.

---

### ADVOSYNTH: A Synthetic Multi-Advocate Dataset for Speaker Identification in Courtroom Scenarios

Authors: Aniket Deroy
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10315v1

As large-scale speech-to-speech models achieve high fidelity, the distinction between synthetic voices in structured environments becomes a vital area of study. This paper introduces Advosynth-500, a specialized dataset comprising 100 synthetic speech files featuring 10 unique advocate identities. Using the Speech Llama Omni model, we simulate five distinct advocate pairs engaged in courtroom arguments. We define specific vocal characteristics for each advocate and present a speaker identification challenge to evaluate the ability of modern systems to map audio files to their respective synthetic origins. Dataset is available at this link-https: //github.com/naturenurtureelite/ADVOSYNTH-500.

---

### Hierarchical Refinement of Universal Multimodal Attacks on Vision-Language Models

Authors: Peng-Fei Zhang, Zi Huang
Published: 2026-01-15
Categories: cs.CV, cs.MM
arXiv: https://arxiv.org/abs/2601.10313v3

Existing adversarial attacks for VLP models are mostly sample-specific, resulting in substantial computational overhead when scaled to large datasets or new scenarios. To overcome this limitation, we propose Hierarchical Refinement Attack (HRA), a multimodal universal attack framework for VLP models. For the image modality, we refine the optimization path by leveraging a temporal hierarchy of historical and estimated future gradients to avoid local minima and stabilize universal perturbation learning. For the text modality, it hierarchically models textual importance by considering both intra- and inter-sentence contributions to identify globally influential words, which are then used as universal text perturbations. Extensive experiments across various downstream tasks, VLP models, and datasets, demonstrate the superior transferability of the proposed universal multimodal attacks.

---

### We Need a More Robust Classifier: Dual Causal Learning Empowers Domain-Incremental Time Series Classification

Authors: Zhipeng Liu, Peibo Duan, Xuan Tang, Haodong Jing, Mingyang Geng, Yongsheng Huang, Jialu Xu, Bin Zhang, Binwu Wang
Published: 2026-01-15
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.10312v1

The World Wide Web thrives on intelligent services that rely on accurate time series classification, which has recently witnessed significant progress driven by advances in deep learning. However, existing studies face challenges in domain incremental learning. In this paper, we propose a lightweight and robust dual-causal disentanglement framework (DualCD) to enhance the robustness of models under domain incremental scenarios, which can be seamlessly integrated into time series classification models. Specifically, DualCD first introduces a temporal feature disentanglement module to capture class-causal features and spurious features. The causal features can offer sufficient predictive power to support the classifier in domain incremental learning settings. To accurately capture these causal features, we further design a dual-causal intervention mechanism to eliminate the influence of both intra-class and inter-class confounding features. This mechanism constructs variant samples by combining the current class's causal features with intra-class spurious features and with causal features from other classes. The causal intervention loss encourages the model to accurately predict the labels of these variant samples based solely on the causal features. Extensive experiments on multiple datasets and models demonstrate that DualCD effectively improves performance in domain incremental scenarios. We summarize our rich experiments into a comprehensive benchmark to facilitate research in domain incremental time series classification.

---

### Multilinguality as Sense Adaptation

Authors: Jan Christian Blaise Cruz, David Ifeoluwa Adelani, Alham Fikri Aji
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10310v1

We approach multilinguality as sense adaptation: aligning latent meaning representations across languages rather than relying solely on shared parameters and scale. In this paper, we introduce SENse-based Symmetric Interlingual Alignment (SENSIA), which adapts a Backpack language model from one language to another by explicitly aligning sense-level mixtures and contextual representations on parallel data, while jointly training a target-language language modeling loss to preserve fluency. Across benchmarks on four typologically diverse languages, SENSIA generally outperforms comparable multilingual alignment methods and achieves competitive accuracy against monolingual from-scratch baselines while using 2-4x less target-language data. Analyses of learned sense geometry indicate that local sense topology and global structure relative to English are largely preserved, and ablations show that the method is robust in terms of design and scale.

---

### The Straight and Narrow: Do LLMs Possess an Internal Moral Path?

Authors: Luoming Hu, Jingjie Zeng, Liang Yang, Hongfei Lin
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10307v1

Enhancing the moral alignment of Large Language Models (LLMs) is a critical challenge in AI safety. Current alignment techniques often act as superficial guardrails, leaving the intrinsic moral representations of LLMs largely untouched. In this paper, we bridge this gap by leveraging Moral Foundations Theory (MFT) to map and manipulate the fine-grained moral landscape of LLMs. Through cross-lingual linear probing, we validate the shared nature of moral representations in middle layers and uncover a shared yet different moral subspace between English and Chinese. Building upon this, we extract steerable Moral Vectors and successfully validate their efficacy at both internal and behavioral levels. Leveraging the high generalizability of morality, we propose Adaptive Moral Fusion (AMF), a dynamic inference-time intervention that synergizes probe detection with vector injection to tackle the safety-helpfulness trade-off. Empirical results confirm that our approach acts as a targeted intrinsic defense, effectively reducing incorrect refusals on benign queries while minimizing jailbreak success rates compared to standard baselines.

---

### Evidence-Augmented Policy Optimization with Reward Co-Evolution for Long-Context Reasoning

Authors: Xin Guan, Zijian Li, Shen Huang, Pengjun Xie, Jingren Zhou, Jiuxin Cao
Published: 2026-01-15
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2601.10306v2

While Reinforcement Learning (RL) has advanced LLM reasoning, applying it to long-context scenarios is hindered by sparsity of outcome rewards. This limitation fails to penalize ungrounded "lucky guesses," leaving the critical process of needle-in-a-haystack evidence retrieval largely unsupervised. To address this, we propose EAPO (Evidence-Augmented Policy Optimization). We first establish the Evidence-Augmented Reasoning paradigm, validating via Tree-Structured Evidence Sampling that precise evidence extraction is the decisive bottleneck for long-context reasoning. Guided by this insight, EAPO introduces a specialized RL algorithm where a reward model computes a Group-Relative Evidence Reward, providing dense process supervision to explicitly improve evidence quality. To sustain accurate supervision throughout training, we further incorporate an Adaptive Reward-Policy Co-Evolution mechanism. This mechanism iteratively refines the reward model using outcome-consistent rollouts, sharpening its discriminative capability to ensure precise process guidance. Comprehensive evaluations across eight benchmarks demonstrate that EAPO significantly enhances long-context reasoning performance compared to SOTA baselines.

---

### DanQing: An Up-to-Date Large-Scale Chinese Vision-Language Pre-training Dataset

Authors: Hengyu Shen, Tiancheng Gu, Bin Qin, Lan Wu, Yuling Wu, Shuo Tan, Zelong Sun, Jun Wang, Nan Wu, Xiang An, Weidong Cai, Ziyong Feng et al.
Published: 2026-01-15
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2601.10305v3

Vision-Language Pre-training (VLP) models have achieved remarkable success by leveraging large-scale image-text pairs. While English-centric models like CLIP and SigLIP benefit from massive datasets (e.g., LAION-400M), the development of Chinese VLP remains bottlenecked by the lack of high-quality, large-scale open-source data. In this paper, we present DanQing, a large-scale Chinese cross-modal dataset containing 100 million high-quality image-text pairs curated from Common Crawl. To ensure superior data quality, we develop an effective systematic pipeline comprising data source selection, text refinement, visual diversification, and cross-modal cross-batch filtering, thereby effectively mitigating the intrinsic noise prevalent in web data. Notably, DanQing incorporates data from 2024-2025, enabling models to capture contemporary semantic trends and emerging concepts. Extensive experiments via continued pretraining of SigLIP2 models demonstrate that DanQing consistently outperforms existing Chinese datasets across diverse downstream tasks, including zero-shot classification, cross-modal retrieval, and Chinese-centric large multimodal model tasks. Furthermore, in-depth analysis of DanQing reveals that it exhibits a more balanced semantic distribution and superior scaling capability compared to existing datasets. To facilitate further research in Chinese vision-language pre-training, we will open-source the DanQing dataset under the Creative Common CC-BY-NC 4.0 license.

---

### SPIKE: Sparse Koopman Regularization for Physics-Informed Neural Networks

Authors: Jose Marie Antonio Miñoza
Published: 2026-01-15
Categories: cs.LG, cs.AI, cs.CE, math.DS, math.NA
arXiv: https://arxiv.org/abs/2601.10282v2

Physics-Informed Neural Networks (PINNs) provide a mesh-free approach for solving differential equations by embedding physical constraints into neural network training. However, PINNs tend to overfit within the training domain, leading to poor generalization when extrapolating beyond trained spatiotemporal regions. This work presents SPIKE (Sparse Physics-Informed Koopman-Enhanced), a framework that regularizes PINNs with continuous-time Koopman operators to learn parsimonious dynamics representations. By enforcing linear dynamics $dz/dt = Az$ in a learned observable space, both PIKE (without explicit sparsity) and SPIKE (with L1 regularization on $A$) learn sparse generator matrices, embodying the parsimony principle that complex dynamics admit low-dimensional structure. Experiments across parabolic, hyperbolic, dispersive, and stiff PDEs, including fluid dynamics (Navier-Stokes) and chaotic ODEs (Lorenz), demonstrate consistent improvements in temporal extrapolation, spatial generalization, and long-term prediction accuracy. The continuous-time formulation with matrix exponential integration provides unconditional stability for stiff systems while avoiding diagonal dominance issues inherent in discrete-time Koopman operators.

---

### Queueing-Aware Optimization of Reasoning Tokens for Accuracy-Latency Trade-offs in LLM Servers

Authors: Emre Ozbas, Melih Bastopcu
Published: 2026-01-15
Categories: cs.LG, cs.AI, cs.IT, cs.NI, math.OC
arXiv: https://arxiv.org/abs/2601.10274v1

We consider a single large language model (LLM) server that serves a heterogeneous stream of queries belonging to $N$ distinct task types. Queries arrive according to a Poisson process, and each type occurs with a known prior probability. For each task type, the server allocates a fixed number of internal thinking tokens, which determines the computational effort devoted to that query. The token allocation induces an accuracy-latency trade-off: the service time follows an approximately affine function of the allocated tokens, while the probability of a correct response exhibits diminishing returns. Under a first-in, first-out (FIFO) service discipline, the system operates as an $M/G/1$ queue, and the mean system time depends on the first and second moments of the resulting service-time distribution. We formulate a constrained optimization problem that maximizes a weighted average accuracy objective penalized by the mean system time, subject to architectural token-budget constraints and queue-stability conditions. The objective function is shown to be strictly concave over the stability region, which ensures existence and uniqueness of the optimal token allocation. The first-order optimality conditions yield a coupled projected fixed-point characterization of the optimum, together with an iterative solution and an explicit sufficient condition for contraction. Moreover, a projected gradient method with a computable global step-size bound is developed to guarantee convergence beyond the contractive regime. Finally, integer-valued token allocations are attained via rounding of the continuous solution, and the resulting performance loss is evaluated in simulation results.

---

### MoST: Mixing Speech and Text with Modality-Aware Mixture of Experts

Authors: Yuxuan Lou, Kai Yang, Yang You
Published: 2026-01-15
Categories: cs.CL, cs.AI, cs.LG, cs.SD
arXiv: https://arxiv.org/abs/2601.10272v1

We present MoST (Mixture of Speech and Text), a novel multimodal large language model that seamlessly integrates speech and text processing through our proposed Modality-Aware Mixture of Experts (MAMoE) architecture. While current multimodal models typically process diverse modality representations with identical parameters, disregarding their inherent representational differences, we introduce specialized routing pathways that direct tokens to modality-appropriate experts based on input type. MAMoE simultaneously enhances modality-specific learning and cross-modal understanding through two complementary components: modality-specific expert groups that capture domain-specific patterns and shared experts that facilitate information transfer between modalities. Building on this architecture, we develop an efficient transformation pipeline that adapts the pretrained MoE language model through strategic post-training on ASR and TTS datasets, followed by fine-tuning with a carefully curated speech-text instruction dataset. A key feature of this pipeline is that it relies exclusively on fully accessible, open-source datasets to achieve strong performance and data efficiency. Comprehensive evaluations across ASR, TTS, audio language modeling, and spoken question answering benchmarks show that MoST consistently outperforms existing models of comparable parameter counts. Our ablation studies confirm that the modality-specific routing mechanism and shared experts design significantly contribute to performance gains across all tested domains. To our knowledge, MoST represents the first fully open-source speech-text LLM built on a Mixture of Experts architecture. \footnote{We release MoST model, training code, inference code, and training data at https://github.com/NUS-HPC-AI-Lab/MoST

---

### Early Fault Detection on CMAPSS with Unsupervised LSTM Autoencoders

Authors: P. Sánchez, K. Reyes, B. Radu, E. Fernández
Published: 2026-01-15
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.10269v1

This paper introduces an unsupervised health-monitoring framework for turbofan engines that does not require run-to-failure labels. First, operating-condition effects in NASA CMAPSS sensor streams are removed via regression-based normalisation; then a Long Short-Term Memory (LSTM) autoencoder is trained only on the healthy portion of each trajectory. Persistent reconstruction error, estimated using an adaptive data-driven threshold, triggers real-time alerts without hand-tuned rules. Benchmark results show high recall and low false-alarm rates across multiple operating regimes, demonstrating that the method can be deployed quickly, scale to diverse fleets, and serve as a complementary early-warning layer to Remaining Useful Life models.

---

### In-Context Source and Channel Coding

Authors: Ziqiong Wang, Tianqi Ren, Rongpeng Li, Zhifeng Zhao, Honggang Zhang
Published: 2026-01-15
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.10267v2

Separate Source-Channel Coding (SSCC) remains attractive for text transmission due to its modularity and compatibility with mature entropy coders and powerful channel codes. However, SSCC often suffers from a pronounced cliff effect in low Signal-to-Noise Ratio (SNR) regimes, where residual bit errors after channel decoding can catastrophically break lossless source decoding, especially for Arithmetic Coding (AC) driven by Large Language Models (LLMs). This paper proposes a receiver-side In-Context Decoding (ICD) framework that enhances SSCC robustness without modifying the transmitter. ICD leverages an Error Correction Code Transformer (ECCT) to obtain bit-wise reliability for the decoded information bits. Based on the context-consistent bitstream, ICD constructs a confidence-ranked candidate pool via reliability-guided bit flipping, samples a compact yet diverse subset of candidates, and applies an LLM-based arithmetic decoder to obtain both reconstructions and sequence-level log-likelihoods. A reliability-likelihood fusion rule then selects the final output. We further provide theoretical guarantees on the stability and convergence of the proposed sampling procedure. Extensive experiments over Additive White Gaussian Noise (AWGN) and Rayleigh fading channels demonstrate consistent gains compared with conventional SSCC baselines and representative Joint Source-Channel Coding (JSCC) schemes.

---

### Measuring Affinity between Attention-Head Weight Subspaces via the Projection Kernel

Authors: Hiroaki Yamagiwa, Yusuke Takase, Hidetoshi Shimodaira
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10266v1

Understanding relationships between attention heads is essential for interpreting the internal structure of Transformers, yet existing metrics do not capture this structure well. We focus on the subspaces spanned by attention-head weight matrices and quantify head-to-head relationships using the Projection Kernel (PK), a principal-angle-based measure of subspace similarity. Experiments show that PK reproduces known head-to-head interactions on the IOI task more clearly than prior metrics such as the Composition Score. We further introduce a framework to quantify the informativeness of PK distributions by comparing them with a reference distribution derived from random orthogonal subspaces. As an application, we analyze a directed graph constructed from PK and show that, in GPT2-small, L4H7 acts as a hub by functioning as an identity head.

---

### Sim2Real Deep Transfer for Per-Device CFO Calibration

Authors: Jingze Zheng, Zhiguo Shi, Shibo He, Chaojie Gu
Published: 2026-01-15
Categories: eess.SP, cs.LG
arXiv: https://arxiv.org/abs/2601.10264v1

Carrier Frequency Offset (CFO) estimation in Orthogonal Frequency Division Multiplexing (OFDM) systems faces significant performance degradation across heterogeneous software-defined radio (SDR) platforms due to uncalibrated hardware impairments. Existing deep neural network (DNN)-based approaches lack device-level adaptation, limiting their practical deployment. This paper proposes a Sim2Real transfer learning framework for per-device CFO calibration, combining simulation-driven pretraining with lightweight receiver adaptation. A backbone DNN is pre-trained on synthetic OFDM signals incorporating parametric hardware distortions (e.g., phase noise, IQ imbalance), enabling generalized feature learning without costly cross-device data collection. Subsequently, only the regression layers are fine-tuned using $1,000$ real frames per target device, preserving hardware-agnostic knowledge while adapting to device-specific impairments. Experiments across three SDR families (USRP B210, USRP N210, HackRF One) achieve $30\times$ BER reduction compared to conventional CP-based methods under indoor multipath conditions. The framework bridges the simulation-to-reality gap for robust CFO estimation, enabling cost-effective deployment in heterogeneous wireless systems.

---

### An Ensemble of Evolutionary Algorithms With Both Crisscross Search and Sparrow Search for Processing Inferior Individuals

Authors: Mingxuan Du, Tingzhang Luo, Ziyang Wang, Chengjun Li
Published: 2026-01-15
Categories: cs.NE
arXiv: https://arxiv.org/abs/2601.10263v1

In the field of artificial intelligence, real parameter single objective optimization is an important direction. Both the Differential Evolution (DE) and the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) demonstrate good performance for real parameter single objective optimization. Nevertheless, there exist other types of evolutionary algorithm for the purpose. In recent years, researchers begin to study long-term search. EA4eig - an ensemble of three DE variants and CMA-ES - performs well for long-term search. In this paper, we introduce two types of evolutionary algorithm proposed recently - crisscross search and sparrow search - into EA4eig as secondary evolutionary algorithms to process inferior individuals. Thus, EA4eigCS is obtained. In our ensemble, the secondary evolutionary algorithms are expected to vary distribution of the population for breaking stagnation. Experimental results show that our EA4eigCS outperforms EA4eig and is competitive when compared with state-of-the-art algorithms. Code and supplementary material are available at:https://anonymous.4open.science/r/EA4eigCS-2A43.

---

### Untangling Input Language from Reasoning Language: A Diagnostic Framework for Cross-Lingual Moral Alignment in LLMs

Authors: Nan Li, Bo Kang, Tijl De Bie
Published: 2026-01-15
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2601.10257v2

When LLMs judge moral dilemmas, do they reach different conclusions in different languages, and if so, why? Two factors could drive such differences: the language of the dilemma itself, or the language in which the model reasons. Standard evaluation conflates these by testing only matched conditions (e.g., English dilemma with English reasoning). We introduce a methodology that separately manipulates each factor, covering also mismatched conditions (e.g., English dilemma with Chinese reasoning), enabling decomposition of their contributions. To study \emph{what} changes, we propose an approach to interpret the moral judgments in terms of Moral Foundations Theory. As a side result, we identify evidence for splitting the Authority dimension into a family-related and an institutional dimension. Applying this methodology to English-Chinese moral judgment with 13 LLMs, we demonstrate its diagnostic power: (1) the framework isolates reasoning-language effects as contributing twice the variance of input-language effects; (2) it detects context-dependency in nearly half of models that standard evaluation misses; and (3) a diagnostic taxonomy translates these patterns into deployment guidance. We release our code and datasets at https://anonymous.4open.science/r/CrossCulturalMoralJudgement.

---

### NoReGeo: Non-Reasoning Geometry Benchmark

Authors: Irina Abdullaeva, Anton Vasiliuk, Elizaveta Goncharova, Temurbek Rahmatullaev, Zagorulko Ivan, Maxim Kurkin, Andrey Kuznetsov
Published: 2026-01-15
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.10254v1

We present NoReGeo, a novel benchmark designed to evaluate the intrinsic geometric understanding of large language models (LLMs) without relying on reasoning or algebraic computation. Unlike existing benchmarks that primarily assess models' proficiency in reasoning-based geometry-where solutions are derived using algebraic methods-NoReGeo focuses on evaluating whether LLMs can inherently encode spatial relationships and recognize geometric properties directly. Our benchmark comprises 2,500 trivial geometric problems spanning 25 categories, each carefully crafted to be solvable purely through native geometric understanding, assuming known object locations. We assess a range of state-of-the-art models on NoReGeo, including frontier models like GPT-4, observing that even the most advanced systems achieve an overall maximum of 65% accuracy in binary classification tasks. Further, our ablation experiments demonstrate that such geometric understanding does not emerge through fine-tuning alone, indicating that effective training for geometric comprehension requires a specialized approach from the outset. Our findings highlight a significant gap in current LLMs' ability to natively grasp geometric concepts, providing a foundation for future research toward models with true geometric cognition.

---

### X-SAM: Boosting Sharpness-Aware Minimization with Dominant-Eigenvector Gradient Correction

Authors: Hongru Duan, Yongle Chen, Lei Guan
Published: 2026-01-15
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.10251v1

Sharpness-Aware Minimization (SAM) aims to improve generalization by minimizing a worst-case perturbed loss over a small neighborhood of model parameters. However, during training, its optimization behavior does not always align with theoretical expectations, since both sharp and flat regions may yield a small perturbed loss. In such cases, the gradient may still point toward sharp regions, failing to achieve the intended effect of SAM. To address this issue, we investigate SAM from a spectral and geometric perspective: specifically, we utilize the angle between the gradient and the leading eigenvector of the Hessian as a measure of sharpness. Our analysis illustrates that when this angle is less than or equal to ninety degrees, the effect of SAM's sharpness regularization can be weakened. Furthermore, we propose an explicit eigenvector-aligned SAM (X-SAM), which corrects the gradient via orthogonal decomposition along the top eigenvector, enabling more direct and efficient regularization of the Hessian's maximum eigenvalue. We prove X-SAM's convergence and superior generalization, with extensive experimental evaluations confirming both theoretical and practical advantages.

---

### Cell Behavior Video Classification Challenge, a benchmark for computer vision methods in time-lapse microscopy

Authors: Raffaella Fiamma Cabini, Deborah Barkauskas, Guangyu Chen, Zhi-Qi Cheng, David E Cicchetti, Judith Drazba, Rodrigo Fernandez-Gonzalez, Raymond Hawkins, Yujia Hu, Jyoti Kini, Charles LeWarne, Xufeng Lin et al.
Published: 2026-01-15
Categories: eess.IV, cs.CV, q-bio.QM
arXiv: https://arxiv.org/abs/2601.10250v1

The classification of microscopy videos capturing complex cellular behaviors is crucial for understanding and quantifying the dynamics of biological processes over time. However, it remains a frontier in computer vision, requiring approaches that effectively model the shape and motion of objects without rigid boundaries, extract hierarchical spatiotemporal features from entire image sequences rather than static frames, and account for multiple objects within the field of view. To this end, we organized the Cell Behavior Video Classification Challenge (CBVCC), benchmarking 35 methods based on three approaches: classification of tracking-derived features, end-to-end deep learning architectures to directly learn spatiotemporal features from the entire video sequence without explicit cell tracking, or ensembling tracking-derived with image-derived features. We discuss the results achieved by the participants and compare the potential and limitations of each approach, serving as a basis to foster the development of computer vision methods for studying cellular dynamics.

---

### coTherapist: A Behavior-Aligned Small Language Model to Support Mental Healthcare Experts

Authors: Prottay Kumar Adhikary, Reena Rawat, Tanmoy Chakraborty
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10246v2

Access to mental healthcare is increasingly strained by workforce shortages and rising demand, motivating the development of intelligent systems that can support mental healthcare experts. We introduce coTherapist, a unified framework utilizing a small language model to emulate core therapeutic competencies through domain-specific fine-tuning, retrieval augmentation, and agentic reasoning. Evaluation on clinical queries demonstrates that coTherapist generates more relevant and clinically grounded responses than contemporary baselines. Using our novel T-BARS rubric and psychometric profiling, we confirm coTherapist exhibits high empathy and therapist-consistent personality traits. Furthermore, human evaluation by domain experts validates that coTherapist delivers accurate, trustworthy, and safe responses. coTherapist was deployed and tested by clinical experts. Collectively, these findings demonstrate that small models can be engineered to exhibit expert-like behavior, offering a scalable pathway for digital mental health tools.

---

### TRIM: Hybrid Inference via Targeted Stepwise Routing in Multi-Step Reasoning Tasks

Authors: Vansh Kapoor, Aman Gupta, Hao Chen, Anurag Beniwal, Jing Huang, Aviral Kumar
Published: 2026-01-15
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2601.10245v2

Multi-step reasoning tasks like mathematical problem solving are vulnerable to cascading failures, where a single incorrect step leads to complete solution breakdown. Current LLM routing methods assign entire queries to one model, treating all reasoning steps as equal. We propose TRIM (Targeted routing in multi-step reasoning tasks), which routes only critical steps$\unicode{x2013}$those likely to derail the solution$\unicode{x2013}$to larger models while letting smaller models handle routine continuations. Our key insight is that targeted step-level interventions can fundamentally transform inference efficiency by confining expensive calls to precisely those steps where stronger models prevent cascading errors. TRIM operates at the step-level: it uses process reward models to identify erroneous steps and makes routing decisions based on step-level uncertainty and budget constraints. We develop several routing strategies within TRIM, ranging from a simple threshold-based policy to more expressive policies that reason about long-horizon accuracy-cost trade-offs and uncertainty in step-level correctness estimates. On MATH-500, even the simplest thresholding strategy surpasses prior routing methods with 5x higher cost efficiency, while more advanced policies match the strong, expensive model's performance using 80% fewer expensive model tokens. On harder benchmarks such as AIME, TRIM achieves up to 6x higher cost efficiency. All methods generalize effectively across math reasoning tasks, demonstrating that step-level difficulty represents fundamental characteristics of reasoning.

---

### Attend to what I say: Highlighting relevant content on slides

Authors: Megha Mariam K M, C. V. Jawahar
Published: 2026-01-15
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.10244v1

Imagine sitting in a presentation, trying to follow the speaker while simultaneously scanning the slides for relevant information. While the entire slide is visible, identifying the relevant regions can be challenging. As you focus on one part of the slide, the speaker moves on to a new sentence, leaving you scrambling to catch up visually. This constant back-and-forth creates a disconnect between what is being said and the most important visual elements, making it hard to absorb key details, especially in fast-paced or content-heavy presentations such as conference talks. This requires an understanding of slides, including text, graphics, and layout. We introduce a method that automatically identifies and highlights the most relevant slide regions based on the speaker's narrative. By analyzing spoken content and matching it with textual or graphical elements in the slides, our approach ensures better synchronization between what listeners hear and what they need to attend to. We explore different ways of solving this problem and assess their success and failure cases. Analyzing multimedia documents is emerging as a key requirement for seamless understanding of content-rich videos, such as educational videos and conference talks, by reducing cognitive strain and improving comprehension. Code and dataset are available at: https://github.com/meghamariamkm2002/Slide_Highlight

---

### Loop as a Bridge: Can Looped Transformers Truly Link Representation Space and Natural Language Outputs?

Authors: Guanxu Chen, Dongrui Liu, Jing Shao
Published: 2026-01-15
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2601.10242v1

Large Language Models (LLMs) often exhibit a gap between their internal knowledge and their explicit linguistic outputs. In this report, we empirically investigate whether Looped Transformers (LTs)--architectures that increase computational depth by iterating shared layers--can bridge this gap by utilizing their iterative nature as a form of introspection. Our experiments reveal that while increasing loop iterations narrows the gap, it is partly driven by a degradation of their internal knowledge carried by representations. Moreover, another empirical analysis suggests that current LTs' ability to perceive representations does not improve across loops; it is only present in the final loop. These results suggest that while LTs offer a promising direction for scaling computational depth, they have yet to achieve the introspection required to truly link representation space and natural language.

---

### Fundamental Limitations of Favorable Privacy-Utility Guarantees for DP-SGD

Authors: Murat Bilgehan Ertan, Marten van Dijk
Published: 2026-01-15
Categories: cs.LG, cs.CR
arXiv: https://arxiv.org/abs/2601.10237v3

Differentially Private Stochastic Gradient Descent (DP-SGD) is the dominant paradigm for private training, but its fundamental limitations under worst-case adversarial privacy definitions remain poorly understood. We analyze DP-SGD in the $f$-differential privacy framework, which characterizes privacy via hypothesis-testing trade-off curves, and study shuffled sampling over a single epoch with $M$ gradient updates. We derive an explicit suboptimal upper bound on the achievable trade-off curve. This result induces a geometric lower bound on the separation $κ$, which is the maximum distance between the mechanism's trade-off curve and the ideal random-guessing line. Because a large separation implies significant adversarial advantage, meaningful privacy requires small $κ$. However, we prove that enforcing a small separation imposes a strict lower bound on the Gaussian noise multiplier $σ$, which directly limits the achievable utility. In particular, under the standard worst-case adversarial model, shuffled DP-SGD must satisfy $$σ\ge \frac{1}{\sqrt{2\ln M}} \quad\text{or}\quad κ\ge\ \frac{1}{\sqrt{8}}\!\left(1-\frac{1}{\sqrt{4π\ln M}}\right),$$ thus cannot simultaneously achieve strong privacy and high utility. Although the noise threshold vanishes asymptotically as $M \to \infty$, the convergence is extremely slow. Even for practically relevant numbers of updates the required noise magnitude remains substantial. We further show that the same limitation extends to Poisson subsampling up to constant factors. Our experiments confirm that the noise levels implied by this bound lead to significant accuracy degradation at realistic training settings, thus showing a bottleneck in DP-SGD under standard worst-case adversarial assumptions.

---

### Who Owns the Text? Design Patterns for Preserving Authorship in AI-Assisted Writing

Authors: Bohan Zhang, Chengke Bu, Paramveer S. Dhillon
Published: 2026-01-15
Categories: cs.HC, cs.AI
arXiv: https://arxiv.org/abs/2601.10236v1

AI writing assistants can reduce effort and improve fluency, but they may also weaken writers' sense of authorship. We study this tension with an ownership-aware co-writing editor that offers on-demand, sentence-level suggestions and tests two common design choices: persona-based coaching and style personalization. In an online study (N=176), participants completed three professional writing tasks: an email without AI help, a proposal with generic AI suggestions, and a cover letter with persona-based coaching, while half received suggestions tailored to a brief sample of their prior writing. Across the two AI-assisted tasks, psychological ownership dropped relative to unassisted writing (about 0.85-1.0 points on a 7-point scale), even as cognitive load decreased (about 0.9 points) and quality ratings stayed broadly similar overall. Persona coaching did not prevent the ownership decline. Style personalization partially restored ownership (about +0.43) and increased AI incorporation in text (+5 percentage points). We distill five design patterns: on-demand initiation, micro-suggestions, voice anchoring, audience scaffolds, and point-of-decision provenance, to guide authorship-preserving writing tools.

---

### GeoSteer: Faithful Chain-of-Thought Steering via Latent Manifold Gradients

Authors: Kentaro Kazama, Daiki Shirafuji, Tatsuhiko Saito
Published: 2026-01-15
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.10229v2

Recent advances in Large Language Models (LLMs) have demonstrated remarkable progress in their reasoning capabilities, such as Chain-of-Thought (CoT). Most approaches rely on CoT rationales. Previous studies have shown that LLMs often generate logically inconsistent reasoning steps even when their final answers are correct. These inconsistencies reduce the reliability of the reasoning process. We propose GeoSteer, a manifold-based framework that improves the quality of intermediate reasoning. The method consists of: (1) constructing a CoT dataset with step-level scores, (2) training a Variational Autoencoder (VAE) model and a quality estimation model to learn a low-dimensional manifold of high-quality CoT trajectories, and (3) steering hidden states of target LLMs toward higher-quality regions in the latent space. This last step enables steering of the hidden states by following gradients along the learned manifold. It facilitates geometrically coherent steering. Evaluation experiments were conducted on the GSM8k dataset using the Qwen3 series. We evaluated performance using two metrics: answer accuracy and overall reasoning quality. GeoSteer improved the accuracy by 0.9 points and enhanced the reasoning quality by 4.5 points on average, compared with those of original LLMs. These results indicate that GeoSteer improves an effective and controllable mechanism for improving the quality of intermediate reasoning in LLMs.

---

### Optimizing Multimodal LLMs for Egocentric Video Understanding: A Solution for the HD-EPIC VQA Challenge

Authors: Sicheng Yang, Yukai Huang, Shitong Sun, Weitong Cai, Jiankang Deng, Jifei Song, Zhensong Zhang
Published: 2026-01-15
Categories: cs.CV, cs.MM, eess.IV
arXiv: https://arxiv.org/abs/2601.10228v1

Multimodal Large Language Models (MLLMs) struggle with complex video QA benchmarks like HD-EPIC VQA due to ambiguous queries/options, poor long-range temporal reasoning, and non-standardized outputs. We propose a framework integrating query/choice pre-processing, domain-specific Qwen2.5-VL fine-tuning, a novel Temporal Chain-of-Thought (T-CoT) prompting for multi-step reasoning, and robust post-processing. This system achieves 41.6% accuracy on HD-EPIC VQA, highlighting the need for holistic pipeline optimization in demanding video understanding. Our code, fine-tuned models are available at https://github.com/YoungSeng/Egocentric-Co-Pilot.

---

### Introduction to optimization methods for training SciML models

Authors: Alena Kopaničáková, Elisa Riccietti
Published: 2026-01-15
Categories: math.NA, cs.AI, math.OC
arXiv: https://arxiv.org/abs/2601.10222v2

Optimization is central to both modern machine learning (ML) and scientific machine learning (SciML), yet the structure of the underlying optimization problems differs substantially across these domains. Classical ML typically relies on stochastic, sample-separable objectives that favor first-order and adaptive gradient methods. In contrast, SciML often involves physics-informed or operator-constrained formulations in which differential operators induce global coupling, stiffness, and strong anisotropy in the loss landscape. As a result, optimization behavior in SciML is governed by the spectral properties of the underlying physical models rather than by data statistics, frequently limiting the effectiveness of standard stochastic methods and motivating deterministic or curvature-aware approaches. This document provides a unified introduction to optimization methods in ML and SciML, emphasizing how problem structure shapes algorithmic choices. We review first- and second-order optimization techniques in both deterministic and stochastic settings, discuss their adaptation to physics-constrained and data-driven SciML models, and illustrate practical strategies through tutorial examples, while highlighting open research directions at the interface of scientific computing and scientific machine learning.

---

### Topo-RAG: Topology-aware retrieval for hybrid text-table documents

Authors: Alex Dantart, Marco Kóvacs-Navarro
Published: 2026-01-15
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.10215v1

In enterprise datasets, documents are rarely pure. They are not just text, nor just numbers; they are a complex amalgam of narrative and structure. Current Retrieval-Augmented Generation (RAG) systems have attempted to address this complexity with a blunt tool: linearization. We convert rich, multidimensional tables into simple Markdown-style text strings, hoping that an embedding model will capture the geometry of a spreadsheet in a single vector. But it has already been shown that this is mathematically insufficient. This work presents Topo-RAG, a framework that challenges the assumption that "everything is text". We propose a dual architecture that respects the topology of the data: we route fluid narrative through traditional dense retrievers, while tabular structures are processed by a Cell-Aware Late Interaction mechanism, preserving their spatial relationships. Evaluated on SEC-25, a synthetic enterprise corpus that mimics real-world complexity, Topo-RAG demonstrates an 18.4% improvement in nDCG@10 on hybrid queries compared to standard linearization approaches. It's not just about searching better; it's about understanding the shape of information.

---

### Beyond Inpainting: Unleash 3D Understanding for Stable Camera-Controlled Video Generation

Authors: Dong-Yu Chen, Yixin Guo, Shuojin Yang, Tai-Jiang Mu, Shi-Min Hu
Published: 2026-01-15
Categories: cs.CV, cs.GR
arXiv: https://arxiv.org/abs/2601.10214v3

Camera control has been extensively studied in conditioned video generation; however, performing precisely altering the camera trajectories while faithfully preserving the video content remains a challenging task. The mainstream approach to achieving precise camera control is warping a 3D representation according to the target trajectory. However, such methods fail to fully leverage the 3D priors of video diffusion models (VDMs) and often fall into the Inpainting Trap, resulting in subject inconsistency and degraded generation quality. To address this problem, we propose DepthDirector, a video re-rendering framework with precise camera controllability. By leveraging the depth video from explicit 3D representation as camera-control guidance, our method can faithfully reproduce the dynamic scene of an input video under novel camera trajectories. Specifically, we design a View-Content Dual-Stream Condition mechanism that injects both the source video and the warped depth sequence rendered under the target viewpoint into the pretrained video generation model. This geometric guidance signal enables VDMs to comprehend camera movements and leverage their 3D understanding capabilities, thereby facilitating precise camera control and consistent content generation. Next, we introduce a lightweight LoRA-based video diffusion adapter to train our framework, fully preserving the knowledge priors of VDMs. Additionally, we construct a large-scale multi-camera synchronized dataset named MultiCam-WarpData using Unreal Engine 5, containing 8K videos across 1K dynamic scenes. Extensive experiments show that DepthDirector outperforms existing methods in both camera controllability and visual quality. Our code and dataset will be publicly available.

---

### PADER: Paillier-based Secure Decentralized Social Recommendation

Authors: Chaochao Chen, Jiaming Qian, Fei Zheng, Yachuan Liu
Published: 2026-01-15
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2601.10212v1

The prevalence of recommendation systems also brings privacy concerns to both the users and the sellers, as centralized platforms collect as much data as possible from them. To keep the data private, we propose PADER: a Paillier-based secure decentralized social recommendation system. In this system, the users and the sellers are nodes in a decentralized network. The training and inference of the recommendation model are carried out securely in a decentralized manner, without the involvement of a centralized platform. To this end, we apply the Paillier cryptosystem to the SoReg (Social Regularization) model, which exploits both user's ratings and social relations. We view the SoReg model as a two-party secure polynomial evaluation problem and observe that the simple bipartite computation may result in poor efficiency. To improve efficiency, we design secure addition and multiplication protocols to support secure computation on any arithmetic circuit, along with an optimal data packing scheme that is suitable for the polynomial computations of real values. Experiment results show that our method only takes about one second to iterate through one user with hundreds of ratings, and training with ~500K ratings for one epoch only takes <3 hours, which shows that the method is practical in real applications. The code is available at https://github.com/GarminQ/PADER.

---