# Research Papers — 2026-01 · bundle page 26 of 189

> Full-text markdown of papers #1251–1300 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/26
[← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/25.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-01/page/27.md)

### XFACTORS: Disentangled Information Bottleneck via Contrastive Supervision

Authors: Alexandre Myara, Nicolas Bourriez, Thomas Boyer, Thomas Lemercier, Ihab Bendidi, Auguste Genovesio
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21688v2

Disentangled representation learning aims to map independent factors of variation to independent representation components. On one hand, purely unsupervised approaches have proven successful on fully disentangled synthetic data, but fail to recover semantic factors from real data without strong inductive biases. On the other hand, supervised approaches are unstable and hard to scale to large attribute sets because they rely on adversarial objectives or auxiliary classifiers. We introduce \textsc{XFactors}, a weakly-supervised VAE framework that disentangles and provides explicit control over a chosen set of factors. Building on the Disentangled Information Bottleneck perspective, we decompose the representation into a residual subspace $\mathcal{S}$ and factor-specific subspaces $\mathcal{T}_1,\ldots,\mathcal{T}_K$ and a residual subspace $\mathcal{S}$. Each target factor is encoded in its assigned $\mathcal{T}_i$ through contrastive supervision: an InfoNCE loss pulls together latents sharing the same factor value and pushes apart mismatched pairs. In parallel, KL regularization imposes a Gaussian structure on both $\mathcal{S}$ and the aggregated factor subspaces, organizing the geometry without additional supervision for non-targeted factors and avoiding adversarial training and classifiers. Across multiple datasets, with constant hyperparameters, \textsc{XFactors} achieves state-of-the-art disentanglement scores and yields consistent qualitative factor alignment in the corresponding subspaces, enabling controlled factor swapping via latent replacement. We further demonstrate that our method scales correctly with increasing latent capacity and evaluate it on the real-world dataset CelebA. Our code is available at \href{https://github.com/ICML26-anon/XFactors}{github.com/ICML26-anon/XFactors}.

---

### Don't be so Stief! Learning KV Cache low-rank approximation over the Stiefel manifold

Authors: Luca Benfenati, Matteo Risso, Andrea Vannozzi, Ahmet Caner Yüzügüler, Lukas Cavigelli, Enrico Macii, Daniele Jahier Pagliari, Alessio Burrello
Published: 2026-01-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21686v2

Key-value (KV) caching enables fast autoregressive decoding but at long contexts becomes a dominant bottleneck in High Bandwidth Memory (HBM) capacity and bandwidth. A common mitigation is to compress cached keys and values by projecting per-head matrices to a lower rank, storing only the projections in the HBM. However, existing post-training approaches typically fit these projections using SVD-style proxy objectives, which may poorly reflect end-to-end reconstruction after softmax, value mixing, and subsequent decoder-layer transformations. For these reasons, we introduce StiefAttention, a post-training KV-cache compression method that learns orthonormal projection bases by directly minimizing decoder-layer output reconstruction error. StiefAttention additionally constructs layer-wise error-rank profiles over candidate ranks, enabling sequential rank allocation under a user-specified KV cache budget. Notably, on Llama3-8B under the same conditions, StiefAttention outperforms EigenAttention by $4.2$ points on C4 perplexity and $8.9$ points on 0-shot MMLU accuracy at iso-compression, yielding lower relative error and higher cosine similarity with respect to the original decoder-layer outputs.

---

### Do Not Waste Your Rollouts: Recycling Search Experience for Efficient Test-Time Scaling

Authors: Xinglin Wang, Jiayi Shi, Shaoxiong Feng, Peiwen Yuan, Yiwei Li, Yueqi Zhang, Chuyi Tan, Ji Zhang, Boyuan Pan, Yao Hu, Kan Li
Published: 2026-01-29
Categories: cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2601.21684v2

Test-Time Scaling enhances the reasoning capabilities of Large Language Models by allocating additional inference compute to broaden the exploration of the solution space. However, existing search strategies typically treat rollouts as disposable samples, where valuable intermediate insights are effectively discarded after each trial. This wasted rollout-level experience leads to substantial computational redundancy, as models repeatedly re-derive discovered conclusions and revisit known dead ends across extensive attempts. To bridge this gap, we propose \textbf{Recycling Search Experience (RSE)}, a self-guided, training-free strategy that turns test-time search from a series of isolated trials into a cumulative, experience-guided process. By actively distilling raw trajectories into a shared experience bank, RSE enables positive recycling of intermediate conclusions to shortcut redundant derivations and negative recycling of failure patterns to prune encountered dead ends. Theoretically, we provide an analysis that formalizes the efficiency gains of RSE over independent sampling in solving complex reasoning tasks. Empirically, extensive experiments on HMMT24, HMMT25, IMO-Bench, and HLE show that RSE consistently outperforms strong baselines under comparable computational budgets, establishing a strong compute-efficiency frontier for test-time scaling.

---

### Can Local Learning Match Self-Supervised Backpropagation?

Authors: Wu S. Zihan, Ariane Delrocq, Wulfram Gerstner, Guillaume Bellec
Published: 2026-01-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21683v2

While end-to-end self-supervised learning with backpropagation (global BP-SSL) has become central for training modern AI systems, theories of local self-supervised learning (local-SSL) have struggled to build functional representations in deep neural networks. To establish a link between global and local rules, we first develop a theory for deep linear networks: we identify conditions for local-SSL algorithms (like Forward-forward or CLAPP) to implement exactly the same weight update as a global BP-SSL. Starting from the theoretical insights, we then develop novel variants of local-SSL algorithms to approximate global BP-SSL in deep non-linear convolutional neural networks. Variants that improve the similarity between gradient updates of local-SSL with those of global BP-SSL also show better performance on image datasets (CIFAR-10, STL-10, and Tiny ImageNet). The best local-SSL rule with the CLAPP loss function matches the performance of a comparable global BP-SSL with InfoNCE or CPC-like loss functions, and improves upon state-of-the-art for local SSL on these benchmarks.

---

### FIT to Forget: Robust Continual Unlearning for Large Language Models

Authors: Xiaoyu Xu, Minxin Du, Kun Fang, Yaxin Xiao, Zhicong Huang, Cheng Hong, Qingqing Ye, Haibo Hu
Published: 2026-01-29
Categories: cs.CL, cs.AI, cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2601.21682v2

While large language models (LLMs) exhibit remarkable capabilities, they increasingly face demands to unlearn memorized privacy-sensitive, copyrighted, or harmful content. Existing unlearning methods primarily focus on \emph{single-shot} scenarios, whereas real-world deletion requests arrive \emph{continually}. Naïvely applying these methods to sequential requests leads to severe utility degradation and catastrophic forgetting. To address this, we propose \fit, a robust continual unlearning framework to process high-volume sequential deletion streams while resisting both catastrophic forgetting and post-unlearning recovery. \fit stabilizes sequential updates through three synergistic mechanisms: redundancy \underline{F}iltering, \underline{I}mportance-aware adaptive algorithm selection, and \underline{T}argeted layer attribution. Furthermore, to facilitate rigorous evaluation, we introduce \textbf{PCH}, a unified benchmark encompassing \textbf{P}ersonal, \textbf{C}opyrighted, and \textbf{H}armful content, alongside two symmetric metrics, Forget Degree (F.D.) and Retain Utility (R.U.), to systematically quantify forgetting-utility trade-offs. Extensive experiments across five LLMs (up to 14B parameters) demonstrate that \fit consistently achieves state-of-the-art unlearning efficacy and utility preservation. Notably, even after hundreds of sequential requests, \fit preserves strong downstream (\eg, GSM8K, MMLU) performance and exhibits superior resilience against relearning and quantization recovery attacks.

---

### LLM4Fluid: Large Language Models as Generalizable Neural Solvers for Fluid Dynamics

Authors: Qisong Xiao, Xinhai Chen, Qinglin Wang, Xiaowei Guo, Binglin Wang, Weifeng Chen, Zhichao Wang, Yunfei Liu, Rui Xia, Hang Zou, Gencheng Liu, Shuai Li et al.
Published: 2026-01-29
Categories: cs.LG, physics.flu-dyn
arXiv: https://arxiv.org/abs/2601.21681v1

Deep learning has emerged as a promising paradigm for spatio-temporal modeling of fluid dynamics. However, existing approaches often suffer from limited generalization to unseen flow conditions and typically require retraining when applied to new scenarios. In this paper, we present LLM4Fluid, a spatio-temporal prediction framework that leverages Large Language Models (LLMs) as generalizable neural solvers for fluid dynamics. The framework first compresses high-dimensional flow fields into a compact latent space via reduced-order modeling enhanced with a physics-informed disentanglement mechanism, effectively mitigating spatial feature entanglement while preserving essential flow structures. A pretrained LLM then serves as a temporal processor, autoregressively predicting the dynamics of physical sequences with time series prompts. To bridge the modality gap between prompts and physical sequences, which can otherwise degrade prediction accuracy, we propose a dedicated modality alignment strategy that resolves representational mismatch and stabilizes long-term prediction. Extensive experiments across diverse flow scenarios demonstrate that LLM4Fluid functions as a robust and generalizable neural solver without retraining, achieving state-of-the-art accuracy while exhibiting powerful zero-shot and in-context learning capabilities. Code and datasets are publicly available at https://github.com/qisongxiao/LLM4Fluid.

---

### Scale-Dependent Semantic Dynamics Revealed by Allan Deviation

Authors: Debayan Dasgupta
Published: 2026-01-29
Categories: cs.CL, physics.data-an
arXiv: https://arxiv.org/abs/2601.21678v1

While language progresses through a sequence of semantic states, the underlying dynamics of this progression remain elusive. Here, we treat the semantic progression of written text as a stochastic trajectory in a high-dimensional state space. We utilize Allan deviation, a tool from precision metrology, to analyze the stability of meaning by treating ordered sentence embeddings as a displacement signal. Our analysis reveals two distinct dynamical regimes: short-time power-law scaling, which differentiates creative literature from technical texts, and a long-time crossover to a stability-limited noise floor. We find that while large language models successfully mimic the local scaling statistics of human text, they exhibit a systematic reduction in their stability horizon. These results establish semantic coherence as a measurable physical property, offering a framework to differentiate the nuanced dynamics of human cognition from the patterns generated by algorithmic models.

---

### Multimodal Visual Surrogate Compression for Alzheimer's Disease Classification

Authors: Dexuan Ding, Ciyuan Peng, Endrowednes Kuantama, Jingcai Guo, Jia Wu, Jian Yang, Amin Beheshti, Ming-Hsuan Yang, Yuankai Qi
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21673v1

High-dimensional structural MRI (sMRI) images are widely used for Alzheimer's Disease (AD) diagnosis. Most existing methods for sMRI representation learning rely on 3D architectures (e.g., 3D CNNs), slice-wise feature extraction with late aggregation, or apply training-free feature extractions using 2D foundation models (e.g., DINO). However, these three paradigms suffer from high computational cost, loss of cross-slice relations, and limited ability to extract discriminative features, respectively. To address these challenges, we propose Multimodal Visual Surrogate Compression (MVSC). It learns to compress and adapt large 3D sMRI volumes into compact 2D features, termed as visual surrogates, which are better aligned with frozen 2D foundation models to extract powerful representations for final AD classification. MVSC has two key components: a Volume Context Encoder that captures global cross-slice context under textual guidance, and an Adaptive Slice Fusion module that aggregates slice-level information in a text-enhanced, patch-wise manner. Extensive experiments on three large-scale Alzheimer's disease benchmarks demonstrate our MVSC performs favourably on both binary and multi-class classification tasks compared against state-of-the-art methods.

---

### Diverse via bounded Agreement: Geometric Regularization for Multimodal Fusion

Authors: Zixuan Xia, Hao Wang, Pengcheng Weng, Yanyu Qian, Yangxin Xu, William Dan, Fei Wang
Published: 2026-01-29
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2601.21670v4

Multimodal fusion is often treated as an optimization-balancing problem, where training signals are adjusted to prevent one modality from dominating the others. However, balanced optimization does not fully determine the geometry of intermediate representations. Supervised multimodal models may still learn low-diversity modality-specific embeddings or allow paired cross-modal observations to drift excessively apart, weakening both unimodal robustness and multimodal fusion. We introduce \regName, a lightweight plug-and-play geometric regularization framework for multimodal representation learning. Rather than enforcing rigid cross-modal alignment, \regName follows a bounded-agreement principle: preserve modality-specific diversity while softly constraining only the portion of paired cross-modal drift that exceeds an admissible agreement band. Operationally, \regName combines a dispersion term that mitigates spectral concentration with an agreement-band anchoring term that controls excessive paired drift, requiring no architectural modification or inference-time overhead. Experiments across audio-visual, image-text, and RF-based benchmarks show that \regName consistently improves multimodal performance and often strengthens unimodal representations. These results suggest that explicitly regulating representation geometry is an effective complement to optimization balancing, and provide evidence that geometry-aware regularization can improve multimodal learning across diverse architectures and domains.

---

### Expected Return Causes Outcome-Level Mode Collapse in Reinforcement Learning and How to Fix It with Inverse Probability Scaling

Authors: Abhijeet Sinha, Sundari Elango, Dianbo Liu
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21669v1

Many reinforcement learning (RL) problems admit multiple terminal solutions of comparable quality, where the goal is not to identify a single optimum but to represent a diverse set of high-quality outcomes. Nevertheless, policies trained by standard expected return maximization routinely collapse onto a small subset of outcomes, a phenomenon commonly attributed to insufficient exploration or weak regularization. We show that this explanation is incomplete: outcome level mode collapse is a structural consequence of the expected-return objective itself. Under idealized learning dynamics, the log-probability ratio between any two outcomes evolves linearly in their reward difference, implying exponential ratio divergence and inevitable collapse independent of the exploration strategy, entropy regularization, or optimization algorithm. We identify the source of this pathology as the probability multiplier inside the expectation and propose a minimal correction: inverse probability scaling, which removes outcome-frequency amplification from the learning signal, fundamentally changes the learning dynamics, and provably yields reward-proportional terminal distributions, preventing collapse in multimodal settings. We instantiate this principle in Group Relative Policy Optimization (GRPO) as a drop-in modification, IPS-GRPO, requiring no auxiliary models or architectural changes. Across different reasoning and molecular generation tasks, IPS-GRPO consistently reduces outcome-level mode collapse while matching or exceeding baseline performance, suggesting that correcting the objective rather than adding exploration heuristics is key to reliable multimodal policy optimization.

---

### From Instruction to Event: Sound-Triggered Mobile Manipulation

Authors: Hao Ju, Shaofei Huang, Hongyu Li, Zihan Ding, Si Liu, Meng Wang, Zhedong Zheng
Published: 2026-01-29
Categories: cs.RO, cs.CV
arXiv: https://arxiv.org/abs/2601.21667v3

Current mobile manipulation research predominantly follows an instruction-driven paradigm, where robots rely on predefined textual commands to execute tasks. However, this setting confines robots to a passive role, limiting robotic autonomy and the ability to react to dynamic environmental events. To address these limitations, we introduce Sound-Triggered Mobile Manipulation (STMM), where robots must actively perceive and interact with sound-emitting objects without explicit action instructions. To support STMM, we develop Habitat-Echo, a simulation platform that integrates sound rendering with physical interaction. We further propose a hierarchical baseline that translates high-level planning into low-level executions, where a task planner predicts a skill chain for policy models to execute sequentially. Experiments indicate the feasibility of perceiving auditory events and executing corresponding physical interactions without explicit instructions. Notably, in challenging multi-event scenarios, the robot successfully isolates the primary sources from overlapping acoustic interference to execute the first interactions, and subsequently proceeds to manipulate the secondary objects. These new challenges of planning-to-execution position STMM as a measurable research direction.The code and datasets will be released upon acceptance.

---

### SONIC-O1: A Real-World Benchmark for Evaluating Multimodal Large Language Models on Audio-Video Understanding

Authors: Ahmed Y. Radwan, Christos Emmanouilidis, Hina Tabassum, Deval Pandya, Shaina Raza
Published: 2026-01-29
Categories: cs.AI, cs.CV
arXiv: https://arxiv.org/abs/2601.21666v2

Multimodal Large Language Models (MLLMs) are a major focus of recent AI research. However, most prior work focuses on static image understanding, while their ability to process sequential audio-video data remains underexplored. This gap highlights the need for a high-quality benchmark to systematically evaluate MLLM performance in a real-world setting. We introduce SONIC-O1, a comprehensive, fully human-verified benchmark of 60 hours (231 clips) spanning 13 real-world conversational domains with 4,958 annotations and demographic metadata. SONIC-O1 evaluates three capabilities: open-ended summarization, multiple-choice question (MCQ) answering, and temporal localization with supporting rationales (reasoning). Across closed- and open-source models, we find that the MCQ accuracy shows the smallest gap between model families, but the best closed-source model outperforms the best open-source model by 22.6% on temporal localization. We further observe accuracy gaps of up to 21.4% on temporal localization across demographic groups, indicating persistent disparities in model behaviour. SONIC-O1 provides an open evaluation suite for temporally grounded and demographically robust multimodal understanding. SONIC-O1 is publicly available for research: Project page (https://vectorinstitute.github.io/sonic-o1/), Dataset (https://huggingface.co/datasets/vector-institute/sonic-o1), GitHub (https://github.com/vectorinstitute/sonic-o1), Leaderboard (https://huggingface.co/spaces/vector-institute/sonic-o1-leaderboard).

---

### AdaptBPE: From General Purpose to Specialized Tokenizers

Authors: Vijini Liyanage, François Yvon
Published: 2026-01-29
Categories: cs.CL
arXiv: https://arxiv.org/abs/2601.21665v1

Subword tokenization methods, such as Byte-Pair Encoding (BPE), significantly impact the performance and efficiency of large language models (LLMs). The standard approach involves training a general-purpose tokenizer that uniformly processes all textual data during both training and inference. However, the use of a generic set of tokens can incur inefficiencies when applying the model to specific domains or languages. To address this limitation, we propose a post-training adaptation strategy that selectively replaces low-utility tokens with more relevant ones based on their frequency in an adaptation corpus. Our algorithm identifies the token inventory that most effectively encodes the adaptation corpus for a given target vocabulary size. Extensive experiments on generation and classification tasks across multiple languages demonstrate that our adapted tokenizers compress test corpora more effectively than baselines using the same vocabulary size. This method serves as a lightweight adaptation mechanism, akin to a vocabulary fine-tuning process, enabling optimized tokenization for specific domains or tasks. Our code and data are available at https://github.com/vijini/Adapt-BPE.git.

---

### SENDAI: A Hierarchical Sparse-measurement, EfficieNt Data AssImilation Framework

Authors: Xingyue Zhang, Yuxuan Bao, Mars Liyao Gao, J. Nathan Kutz
Published: 2026-01-29
Categories: cs.LG, cs.AI, cs.CE
arXiv: https://arxiv.org/abs/2601.21664v1

Bridging the gap between data-rich training regimes and observation-sparse deployment conditions remains a central challenge in spatiotemporal field reconstruction, particularly when target domains exhibit distributional shifts, heterogeneous structure, and multi-scale dynamics absent from available training data. We present SENDAI, a hierarchical Sparse-measurement, EfficieNt Data AssImilation Framework that reconstructs full spatial states from hyper sparse sensor observations by combining simulation-derived priors with learned discrepancy corrections. We demonstrate the performance on satellite remote sensing, reconstructing MODIS (Moderate Resolution Imaging Spectroradiometer) derived vegetation index fields across six globally distributed sites. Using seasonal periods as a proxy for domain shift, the framework consistently outperforms established baselines that require substantially denser observations -- SENDAI achieves a maximum SSIM improvement of 185% over traditional baselines and a 36% improvement over recent high-frequency-based methods. These gains are particularly pronounced for landscapes with sharp boundaries and sub-seasonal dynamics; more importantly, the framework effectively preserves diagnostically relevant structures -- such as field topologies, land cover discontinuities, and spatial gradients. By yielding corrections that are more structurally and spectrally separable, the reconstructed fields are better suited for downstream inference of indirectly observed variables. The results therefore highlight a lightweight and operationally viable framework for sparse-measurement reconstruction that is applicable to physically grounded inference, resource-limited deployment, and real-time monitor and control.

---

### Few-Shot Domain Adaptation with Temporal References and Static Priors for Glacier Calving Front Delineation

Authors: Marcel Dreier, Nora Gourmelon, Dakota Pyles, Thorsten Seehaus, Matthias H. Braun, Andreas Maier, Vincent Christlein
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21663v1

During benchmarking, the state-of-the-art model for glacier calving front delineation achieves near-human performance. However, when applied in a real-world setting at a novel study site, its delineation accuracy is insufficient for calving front products intended for further scientific analyses. This site represents an out-of-distribution domain for a model trained solely on the benchmark dataset. By employing a few-shot domain adaptation strategy, incorporating spatial static prior knowledge, and including summer reference images in the input time series, the delineation error is reduced from 1131.6 m to 68.7 m without any architectural modifications. These methodological advancements establish a framework for applying deep learning-based calving front segmentation to novel study sites, enabling calving front monitoring on a global scale.

---

### Epistemic Uncertainty Quantification for Pre-trained VLMs via Riemannian Flow Matching

Authors: Li Ju, Mayank Nautiyal, Andreas Hellander, Ekta Vats, Prashant Singh
Published: 2026-01-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21662v2

Vision-Language Models (VLMs) are typically deterministic in nature and lack intrinsic mechanisms to quantify epistemic uncertainty, which reflects the model's lack of knowledge or ignorance of its own representations. We theoretically motivate negative log-density of an embedding as a proxy for the epistemic uncertainty, where low-density regions signify model ignorance. The proposed method REPVLM computes the probability density on the hyperspherical manifold of the VLM embeddings using Riemannian Flow Matching. We empirically demonstrate that REPVLM achieves near-perfect correlation between uncertainty and prediction error, significantly outperforming existing baselines. Beyond classification, we also demonstrate that the model also provides a scalable metric for out-of-distribution detection and automated data curation.

---

### TabClustPFN: A Prior-Fitted Network for Tabular Data Clustering

Authors: Tianqi Zhao, Guanyang Wang, Yan Shuo Tan, Qiong Zhang
Published: 2026-01-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21656v3

Clustering tabular data is a fundamental yet challenging problem due to heterogeneous feature types, diverse data-generating mechanisms, and the absence of transferable inductive biases across datasets. Prior-fitted networks (PFNs) have recently demonstrated strong generalization in supervised tabular learning by amortizing Bayesian inference under a broad synthetic prior. Extending this paradigm to clustering is nontrivial: clustering is unsupervised, admits a combinatorial and permutation-invariant output space, and requires inferring the number of clusters. We introduce TabClustPFN, a prior-fitted network for tabular data clustering that performs amortized Bayesian inference over both cluster assignments and cluster cardinality. Pretrained on synthetic datasets drawn from a flexible clustering prior, TabClustPFN clusters unseen datasets in a single forward pass, without dataset-specific retraining or hyperparameter tuning. The model naturally handles heterogeneous numerical and categorical features and adapts to a wide range of clustering structures. Experiments on synthetic data and curated real-world tabular benchmarks show that TabClustPFN outperforms classical, deep, and amortized clustering baselines, while exhibiting strong robustness in out-of-the-box exploratory settings. Code is available at https://github.com/Tianqi-Zhao/TabClustPFN.

---

### ScholarGym: Benchmarking Large Language Model Capabilities in the Information-Gathering Stage of Deep Research

Authors: Hao Shen, Hang Yang, Zhouhong Gu, Weili Han
Published: 2026-01-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21654v3

Large language models have advanced from single-turn question answering to deep research systems that iteratively decompose research questions, invoke retrieval tools, and synthesize information across multiple rounds. Evaluating such systems typically involves scoring their final research reports holistically, but this end-to-end paradigm tightly couples the language model's decision-making, workflow design, and environmental feedback, precluding decomposable analysis of individual components. We introduce ScholarGym, an evaluation environment that isolates the information-gathering stage of deep research on academic literature. Under a unified workflow, ScholarGym decomposes the research process into three explicit stages -- Query Planning, Tool Invocation, and Relevance Assessment -- and evaluates each against 2,536 expert-annotated queries over a static corpus of 570K papers with deterministic retrieval. Systematic experiments reveal that iterative query decomposition yields 2.9--3.3$\times$ F1 gains over single-query retrieval, models with extended thinking trade recall for precision, and Query Planning quality together with Relevance Assessment constitute dual bottlenecks that separate proprietary from open-source model performance.

---

### Gauge-invariant representation holonomy

Authors: Vasileios Sevetlidis, George Pavlidis
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21653v1

Deep networks learn internal representations whose geometry--how features bend, rotate, and evolve--affects both generalization and robustness. Existing similarity measures such as CKA or SVCCA capture pointwise overlap between activation sets, but miss how representations change along input paths. Two models may appear nearly identical under these metrics yet respond very differently to perturbations or adversarial stress. We introduce representation holonomy, a gauge-invariant statistic that measures this path dependence. Conceptually, holonomy quantifies the "twist" accumulated when features are parallel-transported around a small loop in input space: flat representations yield zero holonomy, while nonzero values reveal hidden curvature. Our estimator fixes gauge through global whitening, aligns neighborhoods using shared subspaces and rotation-only Procrustes, and embeds the result back to the full feature space. We prove invariance to orthogonal (and affine, post-whitening) transformations, establish a linear null for affine layers, and show that holonomy vanishes at small radii. Empirically, holonomy increases with loop radius, separates models that appear similar under CKA, and correlates with adversarial and corruption robustness. It also tracks training dynamics as features form and stabilize. Together, these results position representation holonomy as a practical and scalable diagnostic for probing the geometric structure of learned representations beyond pointwise similarity.

---

### When Life Gives You AI, Will You Turn It Into A Market for Lemons? Understanding How Information Asymmetries About AI System Capabilities Affect Market Outcomes and Adoption

Authors: Alexander Erlei, Federico Cau, Radoslav Georgiev, Sagar Kumar, Kilian Bizer, Ujwal Gadiraju
Published: 2026-01-29
Categories: cs.HC, cs.AI
arXiv: https://arxiv.org/abs/2601.21650v1

AI consumer markets are characterized by severe buyer-supplier market asymmetries. Complex AI systems can appear highly accurate while making costly errors or embedding hidden defects. While there have been regulatory efforts surrounding different forms of disclosure, large information gaps remain. This paper provides the first experimental evidence on the important role of information asymmetries and disclosure designs in shaping user adoption of AI systems. We systematically vary the density of low-quality AI systems and the depth of disclosure requirements in a simulated AI product market to gauge how people react to the risk of accidentally relying on a low-quality AI system. Then, we compare participants' choices to a rational Bayesian model, analyzing the degree to which partial information disclosure can improve AI adoption. Our results underscore the deleterious effects of information asymmetries on AI adoption, but also highlight the potential of partial disclosure designs to improve the overall efficiency of human decision-making.

---

### SWE-Spot: Building Small Repo-Experts with Repository-Centric Learning

Authors: Jinjun Peng, Magnus Saebo, Tianjun Zhong, Yi-Jie Cheng, Junfeng Yang, Baishakhi Ray, Simin Chen, Yangruibo Ding
Published: 2026-01-29
Categories: cs.LG, cs.AI, cs.CL, cs.SE
arXiv: https://arxiv.org/abs/2601.21649v1

The deployment of coding agents in privacy-sensitive and resource-constrained environments drives the demand for capable open-weight Small Language Models (SLMs). However, they suffer from a fundamental capability gap: unlike frontier large models, they lack the inference-time strong generalization to work with complicated, unfamiliar codebases. We identify that the prevailing Task-Centric Learning (TCL) paradigm, which scales exposure across disparate repositories, fails to address this limitation. In response, we propose Repository-Centric Learning (RCL), a paradigm shift that prioritizes vertical repository depth over horizontal task breadth, suggesting SLMs must internalize the "physics" of a target software environment through parametric knowledge acquisition, rather than attempting to recover it via costly inference-time search. Following this new paradigm, we design a four-unit Repository-Centric Experience, transforming static codebases into interactive learning signals, to train SWE-Spot-4B, a family of highly compact models built as repo-specialized experts that breaks established scaling trends, outperforming open-weight models up to larger (e.g., CWM by Meta, Qwen3-Coder-30B) and surpassing/matching efficiency-focused commercial models (e.g., GPT-4.1-mini, GPT-5-nano) across multiple SWE tasks. Further analysis reveals that RCL yields higher training sample efficiency and lower inference costs, emphasizing that for building efficient intelligence, repository mastery is a distinct and necessary dimension that complements general coding capability.

---

### CAF-Mamba: Mamba-Based Cross-Modal Adaptive Attention Fusion for Multimodal Depression Detection

Authors: Bowen Zhou, Marc-André Fiedler, Ayoub Al-Hamadi
Published: 2026-01-29
Categories: cs.CV, cs.CY, cs.HC
arXiv: https://arxiv.org/abs/2601.21648v2

Depression is a prevalent mental health disorder that severely impairs daily functioning and quality of life. While recent deep learning approaches for depression detection have shown promise, most rely on limited feature types, overlook explicit cross-modal interactions, and employ simple concatenation or static weighting for fusion. To overcome these limitations, we propose CAF-Mamba, a novel Mamba-based cross-modal adaptive attention fusion framework. CAF-Mamba not only captures cross-modal interactions explicitly and implicitly, but also dynamically adjusts modality contributions through a modality-wise attention mechanism, enabling more effective multimodal fusion. Experiments on two in-the-wild benchmark datasets, LMVD and D-Vlog, demonstrate that CAF-Mamba consistently outperforms existing methods and achieves state-of-the-art performance. Our code is available at https://github.com/zbw-zhou/CAF-Mamba.

---

### ILRR: Inference-Time Steering Method for Masked Diffusion Language Models

Authors: Eden Avrahami, Eliya Nachmani
Published: 2026-01-29
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2601.21647v2

Discrete Diffusion Language Models (DLMs) offer a promising non-autoregressive alternative for text generation, yet effective mechanisms for inference-time control remain relatively underexplored. Existing approaches include sampling-level guidance or trajectory optimization mechanisms. In this work, we study the paradigm of reference-based latent steering for DLMs. We introduce Iterative Latent Representation Refinement (ILRR), an efficient framework for steering DLMs using a reference text as a high-level semantic blueprint. ILRR extracts and injects reference-derived semantic signals into the evolving activations of the generated sequence, enabling tunable transfer of coarse properties such as sentiment. We further introduce Spatially Modulated Steering, an extension that enables long-form generation to be guided by shorter references by regulating intensity across the sequence. Empirically, we demonstrate that ILRR achieves effective control on LLaDA and MDLM architectures with low computational overhead, requiring only one additional parallel forward pass per denoising step. Under comparable compute budgets, ILRR improves attribute accuracy over baselines by 10% to 60% points. Our results suggest that the iterative, global denoising process makes DLMs a natural substrate for effective sequence-wide activation-level control.

---

### Identifiable Equivariant Networks are Layerwise Equivariant

Authors: Vahid Shahverdi, Giovanni Luca Marchetti, Georg Bökman, Kathlén Kohn
Published: 2026-01-29
Categories: cs.LG, math.CT, math.RT
arXiv: https://arxiv.org/abs/2601.21645v2

We investigate the relation between end-to-end equivariance and layerwise equivariance in deep neural networks. We prove the following: For a network whose end-to-end function is equivariant with respect to group actions on the input and output spaces, there is a parameter choice yielding the same end-to-end function such that its layers are equivariant with respect to some group actions on the latent spaces. Our result assumes that the parameters of the model are identifiable in an appropriate sense. This identifiability property has been established in the literature for a large class of networks, to which our results apply immediately, while it is conjectural for others. The theory we develop is grounded in an abstract formalism, and is therefore architecture-agnostic. Overall, our results provide a mathematical explanation for the emergence of equivariant structures in the weights of neural networks during training -- a phenomenon that is consistently observed in practice.

---

### Seg-MoE: Multi-Resolution Segment-wise Mixture-of-Experts for Time Series Forecasting Transformers

Authors: Evandro S. Ortigossa, Eran Segal
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21641v2

Transformer-based models have recently made significant advances in accurate time-series forecasting, but even these architectures struggle to scale efficiently while capturing long-term temporal dynamics. Mixture-of-Experts (MoE) layers are a proven solution to scaling problems in natural language processing. However, existing MoE approaches for time-series forecasting rely on token-wise routing mechanisms, which may fail to exploit the natural locality and continuity of temporal data. In this work, we introduce Seg-MoE, a sparse MoE design that routes and processes contiguous time-step segments rather than making independent expert decisions. Token segments allow each expert to model intra-segment interactions directly, naturally aligning with inherent temporal patterns. We integrate Seg-MoE layers into a time-series Transformer and evaluate it on multiple multivariate long-term forecasting benchmarks. Seg-MoE consistently achieves state-of-the-art forecasting accuracy across almost all prediction horizons, outperforming both dense Transformers and prior token-wise MoE models. Comprehensive ablation studies confirm that segment-level routing is the key factor driving these gains. Our results show that aligning the MoE routing granularity with the inherent structure of time series provides a powerful, yet previously underexplored, inductive bias, opening new avenues for conditionally sparse architectures in sequential data modeling.

---

### OCRVerse: Towards Holistic OCR in End-to-End Vision-Language Models

Authors: Yufeng Zhong, Lei Chen, Xuanle Zhao, Wenkang Han, Liming Zheng, Jing Huang, Deyang Jiang, Yilin Cao, Lin Ma, Zhixiong Zeng
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21639v2

The development of large vision language models drives the demand for managing, and applying massive amounts of multimodal data, making OCR technology, which extracts information from visual images, increasingly popular. However, existing OCR methods primarily focus on recognizing text elements from images or scanned documents (Text-centric OCR), neglecting the identification of visual elements from visually information-dense image sources (Vision-centric OCR), such as charts, web pages and science plots. In reality, these visually information-dense images are widespread on the internet and have significant real-world application value, such as data visualization and web page analysis. In this technical report, we propose OCRVerse, the first holistic OCR method in end-to-end manner that enables unified text-centric OCR and vision-centric OCR. To this end, we constructe comprehensive data engineering to cover a wide range of text-centric documents, such as newspapers, magazines and books, as well as vision-centric rendered composites, including charts, web pages and scientific plots. Moreover, we propose a two-stage SFT-RL multi-domain training method for OCRVerse. SFT directly mixes cross-domain data to train and establish initial domain knowledge, while RL focuses on designing personalized reward strategies for the characteristics of each domain. Specifically, since different domains require various output formats and expected outputs, we provide sufficient flexibility in the RL stage to customize flexible reward signals for each domain, thereby improving cross-domain fusion and avoiding data conflicts. Experimental results demonstrate the effectiveness of OCRVerse, achieving competitive results across text-centric and vision-centric data types, even comparable to large-scale open-source and closed-source models.

---

### Generative Design of Ship Propellers using Conditional Flow Matching

Authors: Patrick Kruger, Rafael Diaz, Simon Hauschulz, Stefan Harries, Hanno Gottschalk
Published: 2026-01-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21637v1

In this paper, we explore the use of generative artificial intelligence (GenAI) for ship propeller design. While traditional forward machine learning models predict the performance of mechanical components based on given design parameters, GenAI models aim to generate designs that achieve specified performance targets. In particular, we employ conditional flow matching to establish a bidirectional mapping between design parameters and simulated noise that is conditioned on performance labels. This approach enables the generation of multiple valid designs corresponding to the same performance targets by sampling over the noise vector. To support model training, we generate data using a vortex lattice method for numerical simulation and analyze the trade-off between model accuracy and the amount of available data. We further propose data augmentation using pseudo-labels derived from less data-intensive forward surrogate models, which can often improve overall model performance. Finally, we present examples of distinct propeller geometries that exhibit nearly identical performance characteristics, illustrating the versatility and potential of GenAI in engineering design.

---

### Sampling-Free Privacy Accounting for Matrix Mechanisms under Random Allocation

Authors: Jan Schuchardt, Nikita Kalinin
Published: 2026-01-29
Categories: cs.LG, cs.CR, stat.ML
arXiv: https://arxiv.org/abs/2601.21636v3

We study privacy amplification for differentially private model training with matrix factorization under random allocation (also known as the balls-in-bins model). Recent work by Choquette-Choo et al. (2025) proposes a sampling-based Monte Carlo approach to compute amplification parameters in this setting. However, their guarantees either only hold with some high probability or require random abstention by the mechanism. Furthermore, the required number of samples for ensuring $(ε,δ)$-DP is inversely proportional to $δ$. In contrast, we develop sampling-free bounds based on Rényi divergence and conditional composition. The former is facilitated by a dynamic programming formulation to efficiently compute the bounds. The latter complements it by offering stronger privacy guarantees for small $ε$, where Rényi divergence bounds inherently lead to an over-approximation. Our framework applies to arbitrary banded and non-banded matrices. Through numerical comparisons, we demonstrate the efficacy of our approach across a broad range of matrix mechanisms used in research and practice.

---

### RSGround-R1: Rethinking Remote Sensing Visual Grounding through Spatial Reasoning

Authors: Shiqi Huang, Shuting He, Bihan Wen
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21634v1

Remote Sensing Visual Grounding (RSVG) aims to localize target objects in large-scale aerial imagery based on natural language descriptions. Owing to the vast spatial scale and high semantic ambiguity of remote sensing scenes, these descriptions often rely heavily on positional cues, posing unique challenges for Multimodal Large Language Models (MLLMs) in spatial reasoning. To leverage this unique feature, we propose a reasoning-guided, position-aware post-training framework, dubbed \textbf{RSGround-R1}, to progressively enhance spatial understanding. Specifically, we first introduce Chain-of-Thought Supervised Fine-Tuning (CoT-SFT) using synthetically generated RSVG reasoning data to establish explicit position awareness. Reinforcement Fine-Tuning (RFT) is then applied, augmented by our newly designed positional reward that provides continuous and distance-aware guidance toward accurate localization. Moreover, to mitigate incoherent localization behaviors across rollouts, we introduce a spatial consistency guided optimization scheme that dynamically adjusts policy updates based on their spatial coherence, ensuring stable and robust convergence. Extensive experiments on RSVG benchmarks demonstrate superior performance and generalization of our model.

---

### A Tilted Seesaw: Revisiting Autoencoder Trade-off for Controllable Diffusion

Authors: Pu Cao, Yiyang Ma, Feng Zhou, Xuedan Yin, Qing Song, Lu Yang
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21633v1

In latent diffusion models, the autoencoder (AE) is typically expected to balance two capabilities: faithful reconstruction and a generation-friendly latent space (e.g., low gFID). In recent ImageNet-scale AE studies, we observe a systematic bias toward generative metrics in handling this trade-off: reconstruction metrics are increasingly under-reported, and ablation-based AE selection often favors the best-gFID configuration even when reconstruction fidelity degrades. We theoretically analyze why this gFID-dominant preference can appear unproblematic for ImageNet generation, yet becomes risky when scaling to controllable diffusion: AEs can induce condition drift, which limits achievable condition alignment. Meanwhile, we find that reconstruction fidelity, especially instance-level measures, better indicates controllability. We empirically validate the impact of tilted autoencoder evaluation on controllability by studying several recent ImageNet AEs. Using a multi-dimensional condition-drift evaluation protocol reflecting controllable generation tasks, we find that gFID is only weakly predictive of condition preservation, whereas reconstruction-oriented metrics are substantially more aligned. ControlNet experiments further confirm that controllability tracks condition preservation rather than gFID. Overall, our results expose a gap between ImageNet-centric AE evaluation and the requirements of scalable controllable diffusion, offering practical guidance for more reliable benchmarking and model selection.

---

### Reinforcement Learning for Adaptive Composition of Quantum Circuit Optimisation Passes

Authors: Daniel Mills, Ifan Williams, Jacob Swain, Gabriel Matos, Enrico Rinaldi, Alexander Koziell-Pipe
Published: 2026-01-29
Categories: quant-ph, cs.LG
arXiv: https://arxiv.org/abs/2601.21629v1

Many quantum software development kits provide a suite of circuit optimisation passes. These passes have been highly optimised and tested in isolation. However, the order in which they are applied is left to the user, or else defined in general-purpose default pass sequences. While general-purpose sequences miss opportunities for optimisation which are particular to individual circuits, designing pass sequences bespoke to particular circuits requires exceptional knowledge about quantum circuit design and optimisation. Here we propose and demonstrate training a reinforcement learning agent to compose optimisation-pass sequences. In particular the agent's action space consists of passes for two-qubit gate count reduction used in default PyTKET pass sequences. For the circuits in our diverse test set, the (mean, median) fraction of two-qubit gates removed by the agent is $(57.7\%, \ 56.7 \%)$, compared to $(41.8 \%, \ 50.0 \%)$ for the next best default pass sequence.

---

### Noise as a Probe: Membership Inference Attacks on Diffusion Models Leveraging Initial Noise

Authors: Puwei Lian, Yujun Cai, Songze Li, Bingkun Bao
Published: 2026-01-29
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2601.21628v2

Diffusion models have achieved remarkable progress in image generation, but their increasing deployment raises serious concerns about privacy and copyright. In particular, fine-tuned models are highly vulnerable, as they are often fine-tuned on small and private datasets. Membership inference attacks (MIAs) are used to assess privacy risks by determining whether a specific sample was part of a model's training data. Existing MIAs against diffusion models either assume obtaining the intermediate results or require training the shadow model on auxiliary datasets. In this work, we utilized a critical yet overlooked vulnerability: the widely used noise schedules fail to fully eliminate semantic information in the images, resulting in residual semantic signals even at the maximum noise step. We empirically demonstrate that the fine-tuned diffusion model captures hidden correlations between the residual semantics in initial noise and the original images. Building on this insight, we propose a simple yet effective membership inference attack, which injects semantic information into the initial noise and infers membership by analyzing the model's generation result. Extensive experiments demonstrate that the semantic initial noise can strongly reveal membership information, highlighting the vulnerability of diffusion models to MIAs. Code is available at https://github.com/S3IC-Lab/NoiseMIA.

---

### HeRo-Q: A General Framework for Stable Low Bit Quantization via Hessian Conditioning

Authors: Jinhao Zhang, Yunquan Zhang, Zicheng yan, Boyang Zhang, Jun Sun, Daning Cheng
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21626v2

Post Training Quantization (PTQ), a mainstream model compression technique, often leads to the paradoxical 'low error, high loss' phenomenon because it focuses solely on minimizing quantization error. The root cause lies in the Hessian matrix of the LLM loss landscape: a few high curvature directions are extremely sensitive to perturbations. To address this, we propose the Hessian Robust Quantization (HeRo Q) algorithm, which applies a lightweight, learnable rotation-compression matrix to the weight space prior to quantization. This joint framework reshapes the loss landscape by reducing the largest Hessian eigenvalue and reducing its max eigenvalue, thereby significantly enhancing robustness to quantization noise. HeRo-Q requires no architectural modifications, incurs negligible computational overhead, and integrates seamlessly into existing PTQ pipelines. Experiments on Llama and Qwen models show that HeRo Q consistently outperforms state of the art methods including GPTQ, AWQ, and SpinQuant not only achieving superior performance under standard W4A8 settings, but also excelling in the highly challenging W3A16 ultra low bit regime, where it boosts GSM8K accuracy on Llama3 8B to 70.15\% and effectively avoids the logical collapse commonly seen in aggressive quantization.

---

### Training Memory in Deep Neural Networks: Mechanisms, Evidence, and Measurement Gaps

Authors: Vasileios Sevetlidis, George Pavlidis
Published: 2026-01-29
Categories: cs.LG
arXiv: https://arxiv.org/abs/2601.21624v1

Modern deep-learning training is not memoryless. Updates depend on optimizer moments and averaging, data-order policies (random reshuffling vs with-replacement, staged augmentations and replay), the nonconvex path, and auxiliary state (teacher EMA/SWA, contrastive queues, BatchNorm statistics). This survey organizes mechanisms by source, lifetime, and visibility. It introduces seed-paired, function-space causal estimands; portable perturbation primitives (carry/reset of momentum/Adam/EMA/BN, order-window swaps, queue/teacher tweaks); and a reporting checklist with audit artifacts (order hashes, buffer/BN checksums, RNG contracts). The conclusion is a protocol for portable, causal, uncertainty-aware measurement that attributes how much training history matters across models, data, and regimes.

---

### LAMP: Look-Ahead Mixed-Precision Inference of Large Language Models

Authors: Stanislav Budzinskiy, Marian Gloser, Tolunay Yilmaz, Ying Hong Tham, Yuanyi Lin, Wenyi Fang, Fan Wu, Philipp Petersen
Published: 2026-01-29
Categories: cs.LG, math.NA
arXiv: https://arxiv.org/abs/2601.21623v2

Mixed-precision computations are a hallmark of the current stage of AI, driving the progress in large language models towards efficient, locally deployable solutions. This article addresses the floating-point computation of compositionally-rich functions, concentrating on transformer inference. Based on the rounding error analysis of a composition $f(g(\mathrm{x}))$, we provide an adaptive strategy that selects a small subset of components of $g(\mathrm{x})$ to be computed more accurately while all other computations can be carried out with lower accuracy. We then explain how this strategy can be applied to different compositions within a transformer and illustrate its overall effect on transformer inference. We study the effectiveness of this algorithm numerically on GPT-2 models and demonstrate that already very low recomputation rates allow for improvements of up to two orders of magnitude in accuracy.

---

### Similarity of Processing Steps in Vision Model Representations

Authors: Matéo Mahaut, Marco Baroni
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21621v1

Recent literature suggests that the bigger the model, the more likely it is to converge to similar, ``universal'' representations, despite different training objectives, datasets, or modalities. While this literature shows that there is an area where model representations are similar, we study here how vision models might get to those representations -- in particular, do they also converge to the same intermediate steps and operations? We therefore study the processes that lead to convergent representations in different models. First, we quantify distance between different model representations at different stages. We follow the evolution of distances between models throughout processing, identifying the processing steps which are most different between models. We find that while layers at similar positions in different models have the most similar representations, strong differences remain. Classifier models, unlike the others, will discard information about low-level image statistics in their final layers. CNN- and transformer-based models also behave differently, with transformer models applying smoother changes to representations from one layer to the next. These distinctions clarify the level and nature of convergence between model representations, and enables a more qualitative account of the underlying processes in image models.

---

### On the Overscaling Curse of Parallel Thinking: System Efficacy Contradicts Sample Efficiency

Authors: Yiming Wang, Zhuosheng Zhang, Rui Wang
Published: 2026-01-29
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2601.21619v2

Parallel thinking improves LLM reasoning through multi-path sampling and aggregation. In standard evaluations, due to a lack of sample-specific priors, all samples share a global budget chosen to maximize dataset accuracy. However, many samples reach their best accuracy with much smaller budgets, causing low budget utilization. This contradiction between system efficacy and sample efficiency constitutes the Overscaling Curse. In this paper, we first provide a formal analysis of the overscaling curse and quantify its prevalence and severity in real-world systems. To break it, we propose Latent Budget Predictor (LanBo), which probes model latent representations to predict sample-specific optimal budgets. LanBo significantly improves budget utilization while maintaining dataset accuracy. We further integrate LanBo into the full decoding pipeline, inspiring Pre-decoding Budget Adaptation (PreAda), a paradigm that allocates budgets before decoding to preserve decoding-time parallelization. LanBo substantially improves hardware-aware efficiency in latency and memory, demonstrating both its practical value and the promise of LanBo for efficient parallel decoding.

---

### Semantic Content Determines Algorithmic Performance

Authors: Martiño Ríos-García, Nawaf Alampara, Kevin Maik Jablonka
Published: 2026-01-29
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2601.21618v1

Counting should not depend on what is being counted; more generally, any algorithm's behavior should be invariant to the semantic content of its arguments. We introduce WhatCounts to test this property in isolation. Unlike prior work that conflates semantic sensitivity with reasoning complexity or prompt variation, WhatCounts is atomic: count items in an unambiguous, delimited list with no duplicates, distractors, or reasoning steps for different semantic types. Frontier LLMs show over 40% accuracy variation depending solely on what is being counted - cities versus chemicals, names versus symbols. Controlled ablations rule out confounds. The gap is semantic, and it shifts unpredictably with small amounts of unrelated fine-tuning. LLMs do not implement algorithms; they approximate them, and the approximation is argument-dependent. As we show with an agentic example, this has implications beyond counting: any LLM function may carry hidden dependencies on the meaning of its inputs.

---

### PathReasoner-R1: Instilling Structured Reasoning into Pathology Vision-Language Model via Knowledge-Guided Policy Optimization

Authors: Songhan Jiang, Fengchun Liu, Ziyue Wang, Linghan Cai, Yongbing Zhang
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21617v1

Vision-Language Models (VLMs) are advancing computational pathology with superior visual understanding capabilities. However, current systems often reduce diagnosis to directly output conclusions without verifiable evidence-linked reasoning, which severely limits clinical trust and hinders expert error rectification. To address these barriers, we construct PathReasoner, the first large-scale dataset of whole-slide image (WSI) reasoning. Unlike previous work reliant on unverified distillation, we develop a rigorous knowledge-guided generation pipeline. By leveraging medical knowledge graphs, we explicitly align structured pathological findings and clinical reasoning with diagnoses, generating over 20K high-quality instructional samples. Based on the database, we propose PathReasoner-R1, which synergizes trajectory-masked supervised fine-tuning with reasoning-oriented reinforcement learning to instill structured chain-of-thought capabilities. To ensure medical rigor, we engineer a knowledge-aware multi-granular reward function incorporating an Entity Reward mechanism strictly aligned with knowledge graphs. This effectively guides the model to optimize for logical consistency rather than mere outcome matching, thereby enhancing robustness. Extensive experiments demonstrate that PathReasoner-R1 achieves state-of-the-art performance on both PathReasoner and public benchmarks across various image scales, equipping pathology models with transparent, clinically grounded reasoning capabilities. Dataset and code are available at https://github.com/cyclexfy/PathReasoner-R1.

---

### Beyond Parameter Finetuning: Test-Time Representation Refinement for Node Classification

Authors: Jiaxin Zhang, Yiqi Wang, Siwei Wang, Xihong Yang, Yu Shi, Xinwang Liu, En Zhu
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21615v1

Graph Neural Networks frequently exhibit significant performance degradation in the out-of-distribution test scenario. While test-time training (TTT) offers a promising solution, existing Parameter Finetuning (PaFT) paradigm suffer from catastrophic forgetting, hindering their real-world applicability. We propose TTReFT, a novel Test-Time Representation FineTuning framework that transitions the adaptation target from model parameters to latent representations. Specifically, TTReFT achieves this through three key innovations: (1) uncertainty-guided node selection for specific interventions, (2) low-rank representation interventions that preserve pre-trained knowledge, and (3) an intervention-aware masked autoencoder that dynamically adjust masking strategy to accommodate the node selection scheme. Theoretically, we establish guarantees for TTReFT in OOD settings. Empirically, extensive experiments across five benchmark datasets demonstrate that TTReFT achieves consistent and superior performance. Our work establishes representation finetuning as a new paradigm for graph TTT, offering both theoretical grounding and immediate practical utility for real-world deployment.

---

### Representation-Regularized Convolutional Audio Transformer for Audio Understanding

Authors: Bing Han, Chushu Zhou, Yifan Yang, Wei Wang, Chenda Li, Wangyou Zhang, Yanmin Qian
Published: 2026-01-29
Categories: eess.AS, cs.AI, cs.SD
arXiv: https://arxiv.org/abs/2601.21612v1

Bootstrap-based Self-Supervised Learning (SSL) has achieved remarkable progress in audio understanding. However, existing methods typically operate at a single level of granularity, limiting their ability to model the diverse temporal and spectral structures inherent in complex audio signals. Furthermore, bootstrapping representations from scratch is computationally expensive, often requiring extensive training to converge. In this work, we propose the Convolutional Audio Transformer (CAT), a unified framework designed to address these challenges. First, to capture hierarchical audio features, CAT incorporates a Multi-resolution Block that aggregates information across varying granularities. Second, to enhance training efficiency, we introduce a Representation Regularization objective. Drawing inspiration from generative modeling, this auxiliary task guides the student model by aligning its predictions with high-quality semantic representations from frozen, pre-trained external encoders. Experimental results demonstrate that CAT significantly outperforms baselines on audio understanding benchmarks. Notably, it achieves competitive performance on the AudioSet 20k dataset with 5 times faster convergence than existing methods. Codes and checkpoints will be released soon at https://github.com/realzhouchushu/CAT.

---

### Thinking Broad, Acting Fast: Latent Reasoning Distillation from Multi-Perspective Chain-of-Thought for E-Commerce Relevance

Authors: Baopu Qiu, Hao Chen, Yuanrong Wu, Changtong Zan, Chao Wei, Weiru Zhang, Xiaoyi Zeng
Published: 2026-01-29
Categories: cs.IR, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2601.21611v1

Effective relevance modeling is crucial for e-commerce search, as it aligns search results with user intent and enhances customer experience. Recent work has leveraged large language models (LLMs) to address the limitations of traditional relevance models, especially for long-tail and ambiguous queries. By incorporating Chain-of-Thought (CoT) reasoning, these approaches improve both accuracy and interpretability through multi-step reasoning. However, two key limitations remain: (1) most existing approaches rely on single-perspective CoT reasoning, which fails to capture the multifaceted nature of e-commerce relevance (e.g., user intent vs. attribute-level matching vs. business-specific rules); and (2) although CoT-enhanced LLM's offer rich reasoning capabilities, their high inference latency necessitates knowledge distillation for real-time deployment, yet current distillation methods discard the CoT rationale structure at inference, using it as a transient auxiliary signal and forfeiting its reasoning utility. To address these challenges, we propose a novel framework that better exploits CoT semantics throughout the optimization pipeline. Specifically, the teacher model leverages Multi-Perspective CoT (MPCoT) to generate diverse rationales and combines Supervised Fine-Tuning (SFT) with Direct Preference Optimization (DPO) to construct a more robust reasoner. For distillation, we introduce Latent Reasoning Knowledge Distillation (LRKD), which endows a student model with a lightweight inference-time latent reasoning extractor, allowing efficient and low-latency internalization of the LLM's sophisticated reasoning capabilities. Evaluated in offline experiments and online A/B tests on an e-commerce search advertising platform serving tens of millions of users daily, our method delivers significant offline gains, showing clear benefits in both commercial performance and user experience.

---

### WMVLM: Evaluating Diffusion Model Image Watermarking via Vision-Language Models

Authors: Zijin Yang, Yu Sun, Kejiang Chen, Jiawei Zhao, Jun Jiang, Weiming Zhang, Nenghai Yu
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21610v2

Digital watermarking is essential for securing generated images from diffusion models. Accurate watermark evaluation is critical for algorithm development, yet existing methods have significant limitations: they lack a unified framework for both residual and semantic watermarks, provide results without interpretability, neglect comprehensive security considerations, and often use inappropriate metrics for semantic watermarks. To address these gaps, we propose WMVLM, the first unified and interpretable evaluation framework for diffusion model image watermarking via vision-language models (VLMs). We redefine quality and security metrics for each watermark type: residual watermarks are evaluated by artifact strength and erasure resistance, while semantic watermarks are assessed through latent distribution shifts. Moreover, we introduce a three-stage training strategy to progressively enable the model to achieve classification, scoring, and interpretable text generation. Experiments show WMVLM outperforms state-of-the-art VLMs with strong generalization across datasets, diffusion models, and watermarking methods.

---

### RecNet: Self-Evolving Preference Propagation for Agentic Recommender Systems

Authors: Bingqian Li, Xiaolei Wang, Junyi Li, Weitao Li, Long Zhang, Sheng Chen, Wayne Xin Zhao, Ji-Rong Wen
Published: 2026-01-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21609v1

Agentic recommender systems leverage Large Language Models (LLMs) to model complex user behaviors and support personalized decision-making. However, existing methods primarily model preference changes based on explicit user-item interactions, which are sparse, noisy, and unable to reflect the real-time, mutual influences among users and items. To address these limitations, we propose RecNet, a self-evolving preference propagation framework that proactively propagates real-time preference updates across related users and items. RecNet consists of two complementary phases. In the forward phase, the centralized preference routing mechanism leverages router agents to integrate preference updates and dynamically propagate them to the most relevant agents. To ensure accurate and personalized integration of propagated preferences, we further introduce a personalized preference reception mechanism, which combines a message buffer for temporary caching and an optimizable, rule-based filter memory to guide selective preference assimilation based on past experience and interests. In the backward phase, the feedback-driven propagation optimization mechanism simulates a multi-agent reinforcement learning framework, using LLMs for credit assignment, gradient analysis, and module-level optimization, enabling continuous self-evolution of propagation strategies. Extensive experiments on various scenarios demonstrate the effectiveness of RecNet in modeling preference propagation for recommender systems.

---

### Search-Based Risk Feature Discovery in Document Structure Spaces under a Constrained Budget

Authors: Saisubramaniam Gopalakrishnan, Harikrishnan P M, Dagnachew Birru
Published: 2026-01-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21608v1

Enterprise-grade Intelligent Document Processing (IDP) systems support high-stakes workflows across finance, insurance, and healthcare. Early-phase system validation under limited budgets mandates uncovering diverse failure mechanisms, rather than identifying a single worst-case document. We formalize this challenge as a Search-Based Software Testing (SBST) problem, aiming to identify complex interactions between document variables, with the objective to maximize the number of distinct failure types discovered within a fixed evaluation budget. Our methodology operates on a combinatorial space of document configurations, rendering instances of structural \emph{risk features} to induce realistic failure conditions. We benchmark a diverse portfolio of search strategies spanning evolutionary, swarm-based, quality-diversity, learning-based, and quantum under identical budget constraints. Through configuration-level exclusivity, win-rate, and cross-temporal overlap analyses, we show that different solvers consistently uncover failure modes that remain undiscovered by specific alternatives at comparable budgets. Crucially, cross-temporal analysis reveals persistent solver-specific discoveries across all evaluated budgets, with no single strategy exhibiting absolute dominance. While the union of all solvers eventually recovers the observed failure space, reliance on any individual method systematically delays the discovery of important risks. These results demonstrate intrinsic solver complementarity and motivate portfolio-based SBST strategies for robust industrial IDP validation.

---

### Age Matters: Analyzing Age-Related Discussions in App Reviews

Authors: Shashiwadana Nirmania, Garima Sharma, Hourieh Khalajzadeh, Mojtaba Shahin
Published: 2026-01-29
Categories: cs.SE, cs.HC, cs.LG
arXiv: https://arxiv.org/abs/2601.21605v1

In recent years, mobile applications have become indispensable tools for managing various aspects of life. From enhancing productivity to providing personalized entertainment, mobile apps have revolutionized people's daily routines. Despite this rapid growth and popularity, gaps remain in how these apps address the needs of users from different age groups. Users of varying ages face distinct challenges when interacting with mobile apps, from younger users dealing with inappropriate content to older users having difficulty with usability due to age-related vision and cognition impairments. Although there have been initiatives to create age-inclusive apps, a limited understanding of user perspectives on age-related issues may hinder developers from recognizing specific challenges and implementing effective solutions. In this study, we explore age discussions in app reviews to gain insights into how mobile apps should cater to users across different age groups.We manually curated a dataset of 4,163 app reviews from the Google Play Store and identified 1,429 age-related reviews and 2,734 non-age-related reviews. We employed eight machine learning, deep learning, and large language models to automatically detect age discussions, with RoBERTa performing the best, achieving a precision of 92.46%. Additionally, a qualitative analysis of the 1,429 age-related reviews uncovers six dominant themes reflecting user concerns.

---

### Dynamics Reveals Structure: Challenging the Linear Propagation Assumption

Authors: Hoyeon Chang, Bálint Mucsányi, Seong Joon Oh
Published: 2026-01-29
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2601.21601v2

Neural networks adapt through first-order parameter updates, yet it remains unclear whether such updates preserve logical coherence. We investigate the geometric limits of the Linear Propagation Assumption (LPA), the premise that local updates coherently propagate to logical consequences. To formalize this, we adopt relation algebra and study three core operations on relations: negation flips truth values, converse swaps argument order, and composition chains relations. For negation and converse, we prove that guaranteeing direction-agnostic first-order propagation necessitates a tensor factorization separating entity-pair context from relation content. However, for composition, we identify a fundamental obstruction. We show that composition reduces to conjunction, and prove that any conjunction well-defined on linear features must be bilinear. Since bilinearity is incompatible with negation, this forces the feature map to collapse. These results suggest that failures in knowledge editing, the reversal curse, and multi-hop reasoning may stem from common structural limitations inherent to the LPA.

---

### CORE: Collaborative Reasoning via Cross Teaching

Authors: Kshitij Mishra, Mirat Aubakirov, Martin Takac, Nils Lukas, Salem Lahlou
Published: 2026-01-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21600v2

Large language models exhibit complementary reasoning errors: on the same instance, one model may succeed with a particular decomposition while another fails. We propose Collaborative Reasoning (CORE), a training-time collaboration framework that converts peer success into a learning signal via a cross-teaching protocol. Each problem is solved in two stages: a cold round of independent sampling, followed by a contexted rescue round in which models that failed receive hint extracted from a successful peer. CORE optimizes a combined reward that balances (i) correctness, (ii) a lightweight DPP-inspired diversity term to reduce error overlap, and (iii) an explicit rescue bonus for successful recovery. We evaluate CORE across four standard reasoning datasets GSM8K, MATH, AIME, and GPQA. With only 1,000 training examples, a pair of small open source models (3B+4B) reaches Pass@2 of 99.54% on GSM8K and 92.08% on MATH, compared to 82.50% and 74.82% for single-model training. On harder datasets, the 3B+4B pair reaches Pass@2 of 77.34% on GPQA (trained on 348 examples) and 79.65% on AIME (trained on 792 examples), using a training-time budget of at most 1536 context tokens and 3072 generated tokens. Overall, these results show that training-time collaboration can reliably convert model complementarity into large gains without scaling model size.

---

### Beyond Imitation: Reinforcement Learning for Active Latent Planning

Authors: Zhi Zheng, Wee Sun Lee
Published: 2026-01-29
Categories: cs.AI
arXiv: https://arxiv.org/abs/2601.21598v1

Aiming at efficient and dense chain-of-thought (CoT) reasoning, latent reasoning methods fine-tune Large Language Models (LLMs) to substitute discrete language tokens with continuous latent tokens. These methods consume fewer tokens compared to the conventional language CoT reasoning and have the potential to plan in a dense latent space. However, current latent tokens are generally supervised based on imitating language labels. Considering that there can be multiple equivalent but diverse CoT labels for a question, passively imitating an arbitrary one may lead to inferior latent token representations and latent reasoning policies, undermining the potential planning ability and resulting in clear gaps between training and testing. In this work, we emphasize the importance of active planning over the representation space of latent tokens in achieving the optimal latent reasoning policy. So, we propose the \underline{A}c\underline{t}ive Latent \underline{P}lanning method (ATP-Latent), which models the supervision process of latent tokens as a conditional variational auto-encoder (VAE) to obtain a smoother latent space. Moreover, to facilitate the most reasonable latent reasoning policy, ATP-Latent conducts reinforcement learning (RL) with an auxiliary coherence reward, which is calculated based on the consistency between VAE-decoded contents of latent tokens, enabling a guided RL process. In experiments on LLaMA-1B, ATP-Latent demonstrates +4.1\% accuracy and -3.3\% tokens on four benchmarks compared to advanced baselines. Codes are available on https://github.com/zz1358m/ATP-Latent-master.

---

### HydroSense: A Dual-Microcontroller IoT Framework for Real-Time Multi-Parameter Water Quality Monitoring with Edge Processing and Cloud Analytics

Authors: Abdul Hasib, A. S. M. Ahsanul Sarkar Akib, Anish Giri
Published: 2026-01-29
Categories: cs.CV
arXiv: https://arxiv.org/abs/2601.21595v1

The global water crisis necessitates affordable, accurate, and real-time water quality monitoring solutions. Traditional approaches relying on manual sampling or expensive commercial systems fail to address accessibility challenges in resource-constrained environments. This paper presents HydroSense, an innovative Internet of Things framework that integrates six critical water quality parameters including pH, dissolved oxygen (DO), temperature, total dissolved solids (TDS), estimated nitrogen, and water level into a unified monitoring system. HydroSense employs a novel dual-microcontroller architecture, utilizing Arduino Uno for precision analog measurements with five-point calibration algorithms and ESP32 for wireless connectivity, edge processing, and cloud integration. The system implements advanced signal processing techniques including median filtering for TDS measurement, temperature compensation algorithms, and robust error handling. Experimental validation over 90 days demonstrates exceptional performance metrics: pH accuracy of plus or minus 0.08 units across the 0 to 14 range, DO measurement stability within plus or minus 0.2 mg/L, TDS accuracy of plus or minus 1.9 percent across 0 to 1000 ppm, and 99.8 percent cloud data transmission reliability. With a total implementation cost of 32,983 BDT (approximately 300 USD), HydroSense achieves an 85 percent cost reduction compared to commercial systems while providing enhanced connectivity through the Firebase real-time database. This research establishes a new paradigm for accessible environmental monitoring, demonstrating that professional-grade water quality assessment can be achieved through intelligent system architecture and cost-effective component selection.

---