# Research Papers — 2026-01 · bundle page 171 of 189 > Full-text markdown of papers #8501–8550 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/171 [← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/170.md) [Next bundle →](https://aipod.dev/papers/archive/2026-01/page/172.md) ### Multi-scale Graph Autoregressive Modeling: Molecular Property Prediction via Next Token Prediction Authors: Zhuoyang Jiang, Yaosen Min, Peiran Jin, Lei Chen Published: 2026-01-05 Categories: cs.LG, q-bio.QM arXiv: https://arxiv.org/abs/2601.02530v3 We present Connection-Aware Motif Sequencing (CamS), a graph-to-sequence representation that enables decoder-only Transformers to learn molecular graphs via standard next-token prediction (NTP). For molecular property prediction, SMILES-based NTP scales well but lacks explicit topology, whereas graph-native masked modeling captures connectivity but risks disrupting the pivotal chemical details (e.g., activity cliffs). CamS bridges this gap by serializing molecular graphs into structure-rich causal sequences. CamS first mines data-driven connection-aware motifs. It then serializes motifs via scaffold-rooted breadth-first search (BFS) to establish a stable core-to-periphery order. Crucially, CamS enables hierarchical modeling by concatenating sequences from fine to coarse motif scales, allowing the model to condition global scaffolds on dense, uncorrupted local structural evidence. We instantiate CamS-LLaMA by pre-training a vanilla LLaMA backbone on CamS sequences. It achieves state-of-the-art performance on MoleculeNet and the activity-cliff benchmark MoleculeACE, outperforming both SMILES-based language models and strong graph baselines. Interpretability analysis confirms that our multi-scale causal serialization effectively drives attention toward cliff-determining differences. --- ### First Provably Optimal Asynchronous SGD for Homogeneous and Heterogeneous Data Authors: Artavazd Maranjyan Published: 2026-01-05 Categories: math.OC, cs.DC, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2601.02523v1 Artificial intelligence has advanced rapidly through large neural networks trained on massive datasets using thousands of GPUs or TPUs. Such training can occupy entire data centers for weeks and requires enormous computational and energy resources. Yet the optimization algorithms behind these runs have not kept pace. Most large scale training still relies on synchronous methods, where workers must wait for the slowest device, wasting compute and amplifying the effects of hardware and network variability. Removing synchronization seems like a simple fix, but asynchrony introduces staleness, meaning updates computed on outdated models. This makes analysis difficult, especially when delays arise from system level randomness rather than algorithmic choices. As a result, the time complexity of asynchronous methods remains poorly understood. This dissertation develops a rigorous framework for asynchronous first order stochastic optimization, focusing on the core challenge of heterogeneous worker speeds. Within this framework, we show that with proper design, asynchronous SGD can achieve optimal time complexity, matching guarantees previously known only for synchronous methods. Our first contribution, Ringmaster ASGD, attains optimal time complexity in the homogeneous data setting by selectively discarding stale updates. The second, Ringleader ASGD, extends optimality to heterogeneous data, common in federated learning, using a structured gradient table mechanism. Finally, ATA improves resource efficiency by learning worker compute time distributions and allocating tasks adaptively, achieving near optimal wall clock time with less computation. Together, these results establish asynchronous optimization as a theoretically sound and practically efficient foundation for distributed learning, showing that coordination without synchronization can be both feasible and optimal. --- ### CT Scans As Video: Efficient Intracranial Hemorrhage Detection Using Multi-Object Tracking Authors: Amirreza Parvahan, Mohammad Hoseyni, Javad Khoramdel, Amirhossein Nikoofard Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02521v1 Automated analysis of volumetric medical imaging on edge devices is severely constrained by the high memory and computational demands of 3D Convolutional Neural Networks (CNNs). This paper develops a lightweight computer vision framework that reconciles the efficiency of 2D detection with the necessity of 3D context by reformulating volumetric Computer Tomography (CT) data as sequential video streams. This video-viewpoint paradigm is applied to the time-sensitive task of Intracranial Hemorrhage (ICH) detection using the Hemorica dataset. To ensure operational efficiency, we benchmarked multiple generations of the YOLO architecture (v8, v10, v11 and v12) in their Nano configurations, selecting the version with the highest mAP@50 to serve as the slice-level backbone. A ByteTrack algorithm is then introduced to enforce anatomical consistency across the $z$-axis. To address the initialization lag inherent in video trackers, a hybrid inference strategy and a spatiotemporal consistency filter are proposed to distinguish true pathology from transient prediction noise. Experimental results on independent test data demonstrate that the proposed framework serves as a rigorous temporal validator, increasing detection Precision from 0.703 to 0.779 compared to the baseline 2D detector, while maintaining high sensitivity. By approximating 3D contextual reasoning at a fraction of the computational cost, this method provides a scalable solution for real-time patient prioritization in resource-constrained environments, such as mobile stroke units and IoT-enabled remote clinics. --- ### Textual Explanations and Their Evaluations for Reinforcement Learning Policy Authors: Ahmad Terra, Mohit Ahmed, Rafia Inam, Elena Fersman, Martin Törngren Published: 2026-01-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.02514v1 Understanding a Reinforcement Learning (RL) policy is crucial for ensuring that autonomous agents behave according to human expectations. This goal can be achieved using Explainable Reinforcement Learning (XRL) techniques. Although textual explanations are easily understood by humans, ensuring their correctness remains a challenge, and evaluations in state-of-the-art remain limited. We present a novel XRL framework for generating textual explanations, converting them into a set of transparent rules, improving their quality, and evaluating them. Expert's knowledge can be incorporated into this framework, and an automatic predicate generator is also proposed to determine the semantic information of a state. Textual explanations are generated using a Large Language Model (LLM) and a clustering technique to identify frequent conditions. These conditions are then converted into rules to evaluate their properties, fidelity, and performance in the deployed environment. Two refinement techniques are proposed to improve the quality of explanations and reduce conflicting information. Experiments were conducted in three open-source environments to enable reproducibility, and in a telecom use case to evaluate the industrial applicability of the proposed XRL framework. This framework addresses the limitations of an existing method, Autonomous Policy Explanation, and the generated transparent rules can achieve satisfactory performance on certain tasks. This framework also enables a systematic and quantitative evaluation of textual explanations, providing valuable insights for the XRL field. --- ### LLM-Enhanced Reinforcement Learning for Time Series Anomaly Detection Authors: Bahareh Golchin, Banafsheh Rekabdar, Danielle Justo Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02511v1 Detecting anomalies in time series data is crucial for finance, healthcare, sensor networks, and industrial monitoring applications. However, time series anomaly detection often suffers from sparse labels, complex temporal patterns, and costly expert annotation. We propose a unified framework that integrates Large Language Model (LLM)-based potential functions for reward shaping with Reinforcement Learning (RL), Variational Autoencoder (VAE)-enhanced dynamic reward scaling, and active learning with label propagation. An LSTM-based RL agent leverages LLM-derived semantic rewards to guide exploration, while VAE reconstruction errors add unsupervised anomaly signals. Active learning selects the most uncertain samples, and label propagation efficiently expands labeled data. Evaluations on Yahoo-A1 and SMD benchmarks demonstrate that our method achieves state-of-the-art detection accuracy under limited labeling budgets and operates effectively in data-constrained settings. This study highlights the promise of combining LLMs with RL and advanced unsupervised techniques for robust, scalable anomaly detection in real-world applications. --- ### hdlib 2.0: Extending Machine Learning Capabilities of Vector-Symbolic Architectures Authors: Fabio Cumbo, Kabir Dhillon, Daniel Blankenberg Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02509v1 Following the initial publication of hdlib, a Python library for designing Vector-Symbolic Architectures (VSA), we introduce a major extension that significantly enhances its machine learning capabilities. VSA, also known as Hyperdimensional Computing, is a computing paradigm that represents and processes information using high-dimensional vectors. While the first version of hdlib established a robust foundation for creating and manipulating these vectors, this update addresses the growing need for more advanced, data-driven modeling within the VSA framework. Here, we present four extensions: significant enhancements to the existing supervised classification model also enabling feature selection, and a new regression model for predicting continuous variables, a clustering model for unsupervised learning, and a graph-based learning model. Furthermore, we propose the first implementation ever of Quantum Hyperdimensional Computing with quantum-powered arithmetic operations and a new Quantum Machine Learning model for supervised learning. hdlib remains open-source and available on GitHub at https://github.com/cumbof/hdlib under the MIT license, and distributed through the Python Package Index (pip install hdlib) and Conda (conda install -c conda-forge hdlib). Documentation and examples of these new features are available on the official Wiki at https://github.com/cumbof/hdlib/wiki. --- ### Enhancing Debugging Skills with AI-Powered Assistance: A Real-Time Tool for Debugging Support Authors: Elizaveta Artser, Daniil Karol, Anna Potriasaeva, Aleksei Rostovskii, Katsiaryna Dzialets, Ekaterina Koshchenko, Xiaotian Su, April Yi Wang, Anastasiia Birillo Published: 2026-01-05 Categories: cs.SE, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2601.02504v1 Debugging is a crucial skill in programming education and software development, yet it is often overlooked in CS curricula. To address this, we introduce an AI-powered debugging assistant integrated into an IDE. It offers real-time support by analyzing code, suggesting breakpoints, and providing contextual hints. Using RAG with LLMs, program slicing, and custom heuristics, it enhances efficiency by minimizing LLM calls and improving accuracy. A three-level evaluation - technical analysis, UX study, and classroom tests - highlights its potential for teaching debugging. --- ### GEM-Style Constraints for PEFT with Dual Gradient Projection in LoRA Authors: Brian Tekmen, Jason Yin, Qianqian Tong Published: 2026-01-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.02500v1 Full fine-tuning of Large Language Models (LLMs) is computationally costly, motivating Continual Learning (CL) approaches that utilize parameter-efficient adapters. We revisit Gradient Episodic Memory (GEM) within the Low-Rank Adapter (LoRA) subspace and introduce I-GEM: a fixed-budget, GPU-resident dual projected-gradient approximation to GEM's quadratic projection. By constraining non-interference solely within the adapter parameters, I-GEM preserves GEM-like stability with orders-of-magnitude lower mean projection overhead. On a 3-task AG News split with induced domain drift, using GPT-2 (355M) and LoRA ($r=8$), I-GEM matches GEM's average accuracy (within $\sim\!0.04$ pts) and outperforms A-GEM by $\sim\!1.4$ pts. Crucially, it reduces projection time vs.\ GEM by a factor of $\sim\!10^3$. These results suggest that applying GEM constraints in the LoRA subspace is a practical pathway for continual learning at the LLM scale. --- ### Polynomial Convergence of Riemannian Diffusion Models Authors: Xingyu Xu, Ziyi Zhang, Yorie Nakahira, Guannan Qu, Yuejie Chi Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02499v1 Diffusion models have demonstrated remarkable empirical success in the recent years and are considered one of the state-of-the-art generative models in modern AI. These models consist of a forward process, which gradually diffuses the data distribution to a noise distribution spanning the whole space, and a backward process, which inverts this transformation to recover the data distribution from noise. Most of the existing literature assumes that the underlying space is Euclidean. However, in many practical applications, the data are constrained to lie on a submanifold of Euclidean space. Addressing this setting, De Bortoli et al. (2022) introduced Riemannian diffusion models and proved that using an exponentially small step size yields a small sampling error in the Wasserstein distance, provided the data distribution is smooth and strictly positive, and the score estimate is $L_\infty$-accurate. In this paper, we greatly strengthen this theory by establishing that, under $L_2$-accurate score estimate, a {\em polynomially small stepsize} suffices to guarantee small sampling error in the total variation distance, without requiring smoothness or positivity of the data distribution. Our analysis only requires mild and standard curvature assumptions on the underlying manifold. The main ingredients in our analysis are Li-Yau estimate for the log-gradient of heat kernel, and Minakshisundaram-Pleijel parametrix expansion of the perturbed heat equation. Our approach opens the door to a sharper analysis of diffusion models on non-Euclidean spaces. --- ### Variational (Energy-Based) Spectral Learning: A Machine Learning Framework for Solving Partial Differential Equations Authors: M. M. Hammad Published: 2026-01-05 Categories: math.NA, cs.LG arXiv: https://arxiv.org/abs/2601.02492v1 We introduce variational spectral learning (VSL), a machine learning framework for solving partial differential equations (PDEs) that operates directly in the coefficient space of spectral expansions. VSL offers a principled bridge between variational PDE theory, spectral discretization, and contemporary machine learning practice. The core idea is to recast a given PDE \( \mathcal{L}u = f \quad \text{in} \quad Q=Ω\times(0,T), \) together with boundary and initial conditions, into differentiable space-time energies built from strong-form least-squares residuals and weak (Galerkin) formulations. The solution is represented as a finite spectral expansion \( u_N(x,t)=\sum_{n=1}^{N} c_n\,φ_n(x,t), \) where $φ_n$ are tensor-product Chebyshev bases in space and time, with Dirichlet-satisfying spatial modes enforcing homogeneous boundary conditions analytically. This yields a compact linear parameterization in the coefficient vector $\mathbf{c}$, while all PDE complexity is absorbed into the variational energy. We show how to construct strong-form and weak-form space-time functionals, augment them with initial-condition and Tikhonov regularization terms, and minimize the resulting objective with gradient-based optimization. In practice, VSL is implemented in TensorFlow using automatic differentiation and Keras cosine-decay-with-restarts learning-rate schedules, enabling robust optimization of moderately sized coefficient vectors. Numerical experiments on benchmark elliptic and parabolic problems, including one- and two-dimensional Poisson, diffusion, and Burgers-type equations, demonstrate that VSL attains accuracy comparable to classical spectral collocation with Crank-Nicolson time stepping, while providing a differentiable objective suitable for modern optimization tooling. --- ### PatchAlign3D: Local Feature Alignment for Dense 3D Shape Understanding Authors: Souhail Hadgi, Bingchen Gong, Ramana Sundararaman, Emery Pierson, Lei Li, Peter Wonka, Maks Ovsjanikov Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02457v2 Current foundation models for 3D shapes excel at global tasks (retrieval, classification) but transfer poorly to local part-level reasoning. Recent approaches leverage vision and language foundation models to directly solve dense tasks through multi-view renderings and text queries. While promising, these pipelines require expensive inference over multiple renderings, depend heavily on large language-model (LLM) prompt engineering for captions, and fail to exploit the inherent 3D geometry of shapes. We address this gap by introducing an encoder-only 3D model that produces language-aligned patch-level features directly from point clouds. Our pre-training approach builds on existing data engines that generate part-annotated 3D shapes by pairing multi-view SAM regions with VLM captioning. Using this data, we train a point cloud transformer encoder in two stages: (1) distillation of dense 2D features from visual encoders such as DINOv2 into 3D patches, and (2) alignment of these patch embeddings with part-level text embeddings through a multi-positive contrastive objective. Our 3D encoder achieves zero-shot 3D part segmentation with fast single-pass inference without any test-time multi-view rendering, while significantly outperforming previous rendering-based and feed-forward approaches across several 3D part segmentation benchmarks. Project website: https://souhail-hadgi.github.io/patchalign3dsite/ --- ### Diagnostic-Driven Layer-Wise Compensation for Post-Training Quantization of Encoder-Decoder ASR Models Authors: Xinyu Wang, Ziyu Zhao, Yajie Luo, Yihong Wu, Liheng Ma, Jingrui Tian, Lei Ding, Xiao-Wen Chang, Peng Lu Published: 2026-01-05 Categories: cs.SD, cs.CL, eess.AS arXiv: https://arxiv.org/abs/2601.02455v3 Layer-wise post-training quantization reconstructs each layer from inputs already altered by the quantized prefix. QEP compensates for this drift with one model-wide coefficient, conflating the model-level operating point with residual variation across layers. We present FADE, which constructs layer-specific coefficients from normalized round-to-nearest distortion and a heuristic calibrated-solver response. It requires no training or per-model coefficient search and adds no inference-time operation. We evaluate seven Whisper, Moonshine, and Qwen3-ASR models at 3 and 4 bits on four English ASR benchmarks. Across 38 settings, FADE lowers mean word error rate relative to fixed QEP-0.5 in 31, although a development-tuned global coefficient recovers much of this gap. The largest absolute reductions occur in 3-bit settings whose final error remains too high for practical use; these results measure collapse mitigation rather than deployable accuracy. In two lower-WER 3-bit cases, FADE reduces the tuned-global result from 3.67 to 3.10 and from 13.03 to 11.63. Most 4-bit differences from the tuned control are within a descriptive tolerance, and one reverses. Paired reruns and within-seed permutations on an outcome-informed subset support assignment sensitivity in selected cases, but do not estimate a matrix-wide success rate. FADE is therefore a layer-wise alternative when per-model coefficient search is unavailable, not a universal replacement for tuned global compensation. --- ### The Rise of Agentic Testing: Multi-Agent Systems for Robust Software Quality Assurance Authors: Saba Naqvi, Mohammad Baqar, Nawaz Ali Mohammad Published: 2026-01-05 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2601.02454v1 Software testing has progressed toward intelligent automation, yet current AI-based test generators still suffer from static, single-shot outputs that frequently produce invalid, redundant, or non-executable tests due to the lack of execution aware feedback. This paper introduces an agentic multi-model testing framework a closed-loop, self-correcting system in which a Test Generation Agent, an Execution and Analysis Agent, and a Review and Optimization Agent collaboratively generate, execute, analyze, and refine tests until convergence. By using sandboxed execution, detailed failure reporting, and iterative regeneration or patching of failing tests, the framework autonomously improves test quality and expands coverage. Integrated into a CI/CD-compatible pipeline, it leverages reinforcement signals from coverage metrics and execution outcomes to guide refinement. Empirical evaluations on microservice based applications show up to a 60% reduction in invalid tests, 30% coverage improvement, and significantly reduced human effort compared to single-model baselines demonstrating that multi-agent, feedback-driven loops can evolve software testing into an autonomous, continuously learning quality assurance ecosystem for self-healing, high-reliability codebases. --- ### mHC-GNN: Manifold-Constrained Hyper-Connections for Graph Neural Networks Authors: Subhankar Mishra Published: 2026-01-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.02451v2 Graph Neural Networks (GNNs) suffer from over-smoothing in deep architectures and expressiveness bounded by the 1-Weisfeiler-Leman (1-WL) test. We adapt Manifold-Constrained Hyper-Connections, recently proposed for Transformers, to graph neural networks. Our method, \mhcgnn{}, expands node representations across $n$ parallel streams and constrains stream-mixing matrices to the Birkhoff polytope of doubly stochastic matrices via Sinkhorn-Knopp normalization. We prove that \mhcgnn{} mitigates over-smoothing via a layer-wise residual lower bound showing that node-pair differences decay at rate $(1{-}\varepsilon)^L$ (where $\varepsilon$ measures deviation of the mixing matrix from identity), far slower than the standard $(1{-}γ)^L$ collapse rate driven by the spectral gap $γ$. This two-regime analysis, via the protected orthogonal subspace for $L < n$ and the layer-wise contraction for $L \geq n$, provides architecture-agnostic rate guarantees absent from prior methods. With independent random stream initialization, \mhcgnn{} can distinguish graphs beyond 1-WL by maintaining stream diversity across layers via doubly stochastic mixing. Depth experiments spanning 2 to 128 layers reveal that standard GNNs collapse to near-random performance beyond 16 layers, while \mhcgnn{} maintains over 74\% accuracy at 128 layers, with improvements exceeding 50 percentage points at extreme depths. Ablations confirm that manifold constraints are essential: removing them causes up to 82\% performance degradation. Experiments on heterophilic graphs (roman-empire, penn94, genius) and expressiveness benchmarks (EXP) further validate the contribution. Code is available at https://github.com/smishra-lab/mhc-gnn --- ### Don't Mind the Gaps: Implicit Neural Representations for Resolution-Agnostic Retinal OCT Analysis Authors: Bennet Kahrs, Julia Andresen, Fenja Falta, Monty Santarossa, Heinz Handels, Timo Kepp Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02447v3 Routine clinical imaging of the retina using optical coherence tomography (OCT) is performed with large slice spacing, resulting in highly anisotropic images and a sparsely scanned retina. Most learning-based methods circumvent the problems arising from the anisotropy by using 2D approaches rather than performing volumetric analyses. These approaches inherently bear the risk of generating inconsistent results for neighboring B-scans. For example, 2D retinal layer segmentations can have irregular surfaces in 3D. Furthermore, the typically used convolutional neural networks are bound to the resolution of the training data, which prevents their usage for images acquired with a different imaging protocol. Implicit neural representations (INRs) have recently emerged as a tool to store voxelized data as a continuous representation. Using coordinates as input, INRs are resolution-agnostic, which allows them to be applied to anisotropic data. In this paper, we propose two frameworks that make use of this characteristic of INRs for dense 3D analyses of retinal OCT volumes. 1) We perform inter-B-scan interpolation by incorporating additional information from en-face modalities, that help retain relevant structures between B-scans. 2) We create a resolution-agnostic retinal atlas that enables general analysis without strict requirements for the data. Both methods leverage generalizable INRs, improving retinal shape representation through population-based training and allowing predictions for unseen cases. Our resolution-independent frameworks facilitate the analysis of OCT images with large B-scan distances, opening up possibilities for the volumetric evaluation of retinal structures and pathologies. --- ### A Spatio-Temporal Deep Learning Approach For High-Resolution Gridded Monsoon Prediction Authors: Parashjyoti Borah, Sanghamitra Sarkar, Ranjan Phukan Published: 2026-01-05 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.02445v1 The Indian Summer Monsoon (ISM) is a critical climate phenomenon, fundamentally impacting the agriculture, economy, and water security of over a billion people. Traditional long-range forecasting, whether statistical or dynamical, has predominantly focused on predicting a single, spatially-averaged seasonal value, lacking the spatial detail essential for regional-level resource management. To address this gap, we introduce a novel deep learning framework that reframes gridded monsoon prediction as a spatio-temporal computer vision task. We treat multi-variable, pre-monsoon atmospheric and oceanic fields as a sequence of multi-channel images, effectively creating a video-like input tensor. Using 85 years of ERA5 reanalysis data for predictors and IMD rainfall data for targets, we employ a Convolutional Neural Network (CNN)-based architecture to learn the complex mapping from the five-month pre-monsoon period (January-May) to a high-resolution gridded rainfall pattern for the subsequent monsoon season. Our framework successfully produces distinct forecasts for each of the four monsoon months (June-September) as well as the total seasonal average, demonstrating its utility for both intra-seasonal and seasonal outlooks. --- ### VocalBridge: Latent Diffusion-Bridge Purification for Defeating Perturbation-Based Voiceprint Defenses Authors: Maryam Abbasihafshejani, AHM Nazmus Sakib, Murtuza Jadliwala Published: 2026-01-05 Categories: cs.SD, cs.AI, cs.CR, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2601.02444v1 The rapid advancement of speech synthesis technologies, including text-to-speech (TTS) and voice conversion (VC), has intensified security and privacy concerns related to voice cloning. Recent defenses attempt to prevent unauthorized cloning by embedding protective perturbations into speech to obscure speaker identity while maintaining intelligibility. However, adversaries can apply advanced purification techniques to remove these perturbations, recover authentic acoustic characteristics, and regenerate cloneable voices. Despite the growing realism of such attacks, the robustness of existing defenses under adaptive purification remains insufficiently studied. Most existing purification methods are designed to counter adversarial noise in automatic speech recognition (ASR) systems rather than speaker verification or voice cloning pipelines. As a result, they fail to suppress the fine-grained acoustic cues that define speaker identity and are often ineffective against speaker verification attacks (SVA). To address these limitations, we propose Diffusion-Bridge (VocalBridge), a purification framework that learns a latent mapping from perturbed to clean speech in the EnCodec latent space. Using a time-conditioned 1D U-Net with a cosine noise schedule, the model enables efficient, transcript-free purification while preserving speaker-discriminative structure. We further introduce a Whisper-guided phoneme variant that incorporates lightweight temporal guidance without requiring ground-truth transcripts. Experimental results show that our approach consistently outperforms existing purification methods in recovering cloneable voices from protected speech. Our findings demonstrate the fragility of current perturbation-based defenses and highlight the need for more robust protection mechanisms against evolving voice-cloning and speaker verification threats. --- ### Evaluating the Diagnostic Classification Ability of Multimodal Large Language Models: Insights from the Osteoarthritis Initiative Authors: Li Wang, Xi Chen, XiangWen Deng, HuaHui Yi, ZeKun Jiang, Kang Li, Jian Li Published: 2026-01-05 Categories: cs.CV, cs.AI, eess.IV arXiv: https://arxiv.org/abs/2601.02443v1 Multimodal large language models (MLLMs) show promising performance on medical visual question answering (VQA) and report generation, but these generation and explanation abilities do not reliably transfer to disease-specific classification. We evaluated MLLM architectures on knee osteoarthritis (OA) radiograph classification, which remains underrepresented in existing medical MLLM benchmarks, even though knee OA affects an estimated 300 to 400 million people worldwide. Through systematic ablation studies manipulating the vision encoder, the connector, and the large language model (LLM) across diverse training strategies, we measured each component's contribution to diagnostic accuracy. In our classification task, a trained vision encoder alone could outperform full MLLM pipelines in classification accuracy and fine-tuning the LLM provided no meaningful improvement over prompt-based guidance. And LoRA fine-tuning on a small, class-balanced dataset (500 images) gave better results than training on a much larger but class-imbalanced set (5,778 images), indicating that data balance and quality can matter more than raw scale for this task. These findings suggest that for domain-specific medical classification, LLMs are more effective as interpreters and report generators rather than as primary classifiers. Therefore, the MLLM architecture appears less suitable for medical image diagnostic classification tasks that demand high certainty. We recommend prioritizing vision encoder optimization and careful dataset curation when developing clinically applicable systems. --- ### Understanding Pure Textual Reasoning for Blind Image Quality Assessment Authors: Yuan Li, Shin'ya Nishida Published: 2026-01-05 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.02441v2 Textual reasoning has recently been widely adopted in Blind Image Quality Assessment (BIQA). However, it remains unclear how textual information contributes to quality prediction and to what extent text can represent the score-related image contents. This work addresses these questions from an information-flow perspective by comparing existing BIQA models with three paradigms designed to learn the image-text-score relationship: Chain-of-Thought, Self-Consistency, and Autoencoder. Our experiments show that the score prediction performance of the existing model significantly drops when only textual information is used for prediction. Whereas the Chain-of-Thought paradigm introduces little improvement in BIQA performance, the Self-Consistency paradigm significantly reduces the gap between image- and text-conditioned predictions, narrowing the PLCC/SRCC difference to 0.02/0.03. The Autoencoder-like paradigm is less effective in closing the image-text gap, yet it reveals a direction for further optimization. These findings provide insights into how to improve the textual reasoning for BIQA and high-level vision tasks. --- ### Mitigating Long-Tailed Anomaly Score Distributions with Importance-Weighted Loss Authors: Jungi Lee, Jungkwon Kim, Chi Zhang, Sangmin Kim, Kwangsun Yoo, Seok-Joo Byun Published: 2026-01-05 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.02440v1 Anomaly detection is crucial in industrial applications for identifying rare and unseen patterns to ensure system reliability. Traditional models, trained on a single class of normal data, struggle with real-world distributions where normal data exhibit diverse patterns, leading to class imbalance and long-tailed anomaly score distributions (LTD). This imbalance skews model training and degrades detection performance, especially for minority instances. To address this issue, we propose a novel importance-weighted loss designed specifically for anomaly detection. Compared to the previous method for LTD in classification, our method does not require prior knowledge of normal data classes. Instead, we introduce a weighted loss function that incorporates importance sampling to align the distribution of anomaly scores with a target Gaussian, ensuring a balanced representation of normal data. Extensive experiments on three benchmark image datasets and three real-world hyperspectral imaging datasets demonstrate the robustness of our approach in mitigating LTD-induced bias. Our method improves anomaly detection performance by 0.043, highlighting its effectiveness in real-world applications. --- ### WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks Authors: Hao Bai, Alexey Taymanov, Tong Zhang, Aviral Kumar, Spencer Whitehead Published: 2026-01-05 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2601.02439v6 We present WebGym, the largest-to-date open-source environment for training realistic visual web agents. Real websites are non-stationary and diverse, making artificial or small-scale task sets insufficient for robust policy learning. WebGym contains nearly 300,000 tasks with rubric-based evaluations across diverse, real-world websites and difficulty levels. We train agents with a simple reinforcement learning (RL) recipe, which trains on the agent's own interaction traces (rollouts), using task rewards as feedback to guide learning. To enable scaling RL, we speed up sampling of trajectories in WebGym by developing a high-throughput asynchronous rollout system, designed specifically for web agents. Our system achieves a 4-5x rollout speedup compared to naive implementations. Second, we scale the task set breadth, depth, and size, which results in continued performance improvement. Fine-tuning a strong base vision-language model, Qwen-3-VL-8B-Instruct, on WebGym results in an improvement in success rate on an out-of-distribution test set from 26.2% to 42.9%, significantly outperforming agents based on proprietary models such as GPT-4o and GPT-5-Thinking that achieve 27.1% and 29.8%, respectively. This improvement is substantial because our test set consists only of tasks on websites never seen during training, unlike many other prior works on training visual web agents. --- ### Focus on What Matters: Fisher-Guided Adaptive Multimodal Fusion for Vulnerability Detection Authors: Yun Bian, Yi Chen, HaiQuan Wang, ShiHao Li, Zhe Cui Published: 2026-01-05 Categories: cs.SE, cs.AI, cs.CR arXiv: https://arxiv.org/abs/2601.02438v4 Software vulnerability detection can be formulated as a binary classification problem that determines whether a given code snippet contains security defects. Existing multimodal methods typically fuse Natural Code Sequence (NCS) representations extracted by pretrained models with Code Property Graph (CPG) representations extracted by graph neural networks, under the implicit assumption that introducing an additional modality necessarily yields information gain. Through empirical analysis, we demonstrate the limitations of this assumption: pretrained models already encode substantial structural information implicitly, leading to strong overlap between the two modalities; moreover, graph encoders are generally less effective than pretrained language models in feature extraction. As a result, naive fusion not only struggles to obtain complementary signals but can also dilute effective discriminative cues due to noise propagation. To address these challenges, we propose a task-conditioned complementary fusion strategy that uses Fisher information to quantify task relevance, transforming cross-modal interaction from full-spectrum matching into selective fusion within a task-sensitive subspace. Our theoretical analysis shows that, under an isotropic perturbation assumption, this strategy significantly tightens the upper bound on the output error. Based on this insight, we design the TaCCS-DFA framework, which combines online low-rank Fisher subspace estimation with an adaptive gating mechanism to enable efficient task-oriented fusion. Experiments on the BigVul, Devign, and ReVeal benchmarks demonstrate that TaCCS-DFA delivers up to a 6.3-point gain in F1 score with only a 3.4% increase in inference latency, while maintaining low calibration error. --- ### TAP-ViTs: Task-Adaptive Pruning for On-Device Deployment of Vision Transformers Authors: Zhibo Wang, Zuoyuan Zhang, Xiaoyi Pang, Qile Zhang, Xuanyi Hao, Shuguo Zhuo, Peng Sun Published: 2026-01-05 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.02437v1 Vision Transformers (ViTs) have demonstrated strong performance across a wide range of vision tasks, yet their substantial computational and memory demands hinder efficient deployment on resource-constrained mobile and edge devices. Pruning has emerged as a promising direction for reducing ViT complexity. However, existing approaches either (i) produce a single pruned model shared across all devices, ignoring device heterogeneity, or (ii) rely on fine-tuning with device-local data, which is often infeasible due to limited on-device resources and strict privacy constraints. As a result, current methods fall short of enabling task-customized ViT pruning in privacy-preserving mobile computing settings. This paper introduces TAP-ViTs, a novel task-adaptive pruning framework that generates device-specific pruned ViT models without requiring access to any raw local data. Specifically, to infer device-level task characteristics under privacy constraints, we propose a Gaussian Mixture Model (GMM)-based metric dataset construction mechanism. Each device fits a lightweight GMM to approximate its private data distribution and uploads only the GMM parameters. Using these parameters, the cloud selects distribution-consistent samples from public data to construct a task-representative metric dataset for each device. Based on this proxy dataset, we further develop a dual-granularity importance evaluation-based pruning strategy that jointly measures composite neuron importance and adaptive layer importance, enabling fine-grained, task-aware pruning tailored to each device's computational budget. Extensive experiments across multiple ViT backbones and datasets demonstrate that TAP-ViTs consistently outperforms state-of-the-art pruning methods under comparable compression ratios. --- ### Deep Learning Superresolution for 7T Knee MR Imaging: Impact on Image Quality and Diagnostic Performance Authors: Pinzhen Chen, Libo Xu, Boyang Pan, Jing Li, Yuting Wang, Ran Xiong, Xiaoli Gou, Long Qing, Wenjing Hou, Nan-jie Gong, Wei Chen Published: 2026-01-05 Categories: eess.IV, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.02436v1 Background: Deep learning superresolution (SR) may enhance musculoskeletal MR image quality, but its diagnostic value in knee imaging at 7T is unclear. Objectives: To compare image quality and diagnostic performance of SR, low-resolution (LR), and high-resolution (HR) 7T knee MRI. Methods: In this prospective study, 42 participants underwent 7T knee MRI with LR (0.8*0.8*2 mm3) and HR (0.4*0.4*2 mm3) sequences. SR images were generated from LR data using a Hybrid Attention Transformer model. Three radiologists assessed image quality, anatomic conspicuity, and detection of knee pathologies. Arthroscopy served as reference in 10 cases. Results: SR images showed higher overall quality than LR (median score 5 vs 4, P<.001) and lower noise than HR (5 vs 4, P<.001). Visibility of cartilage, menisci, and ligaments was superior in SR and HR compared to LR (P<.001). Detection rates and diagnostic performance (sensitivity, specificity, AUC) for intra-articular pathology were similar across image types (P>=.095). Conclusions: Deep learning superresolution improved subjective image quality in 7T knee MRI but did not increase diagnostic accuracy compared with standard LR imaging. --- ### Physical Transformer Authors: Tao Xu, Zhixin Hu, Li Luo, Momiao Xiong Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02433v1 Digital AI systems spanning large language models, vision models, and generative architectures that operate primarily in symbolic, linguistic, or pixel domains. They have achieved striking progress, but almost all of this progress lives in virtual spaces. These systems transform embeddings and tokens, yet do not themselves touch the world and rarely admit a physical interpretation. In this work we propose a physical transformer that couples modern transformer style computation with geometric representation and physical dynamics. At the micro level, attention heads, and feed-forward blocks are modeled as interacting spins governed by effective Hamiltonians plus non-Hamiltonian bath terms. At the meso level, their aggregated state evolves on a learned Neural Differential Manifold (NDM) under Hamiltonian flows and Hamilton, Jacobi, Bellman (HJB) optimal control, discretized by symplectic layers that approximately preserve geometric and energetic invariants. At the macro level, the model maintains a generative semantic workspace and a two-dimensional information-phase portrait that tracks uncertainty and information gain over a reasoning trajectory. Within this hierarchy, reasoning tasks are formulated as controlled information flows on the manifold, with solutions corresponding to low cost trajectories that satisfy geometric, energetic, and workspace-consistency constraints. On simple toy problems involving numerical integration and dynamical systems, the physical transformer outperforms naive baselines in stability and long-horizon accuracy, highlighting the benefits of respecting underlying geometric and Hamiltonian structure. More broadly, the framework suggests a path toward physical AI that unify digital reasoning with physically grounded manifolds, opening a route to more interpretable and potentially unified models of reasoning, control, and interaction with the real world. --- ### Quantifying Quanvolutional Neural Networks Robustness for Speech in Healthcare Applications Authors: Ha Tran, Bipasha Kashyap, Pubudu N. Pathirana Published: 2026-01-05 Categories: cs.SD, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2601.02432v1 Speech-based machine learning systems are sensitive to noise, complicating reliable deployment in emotion recognition and voice pathology detection. We evaluate the robustness of a hybrid quantum machine learning model, quanvolutional neural networks (QNNs) against classical convolutional neural networks (CNNs) under four acoustic corruptions (Gaussian noise, pitch shift, temporal shift, and speed variation) in a clean-train/corrupted-test regime. Using AVFAD (voice pathology) and TESS (speech emotion), we compare three QNN models (Random, Basic, Strongly) to a simple CNN baseline (CNN-Base), ResNet-18 and VGG-16 using accuracy and corruption metrics (CE, mCE, RCE, RmCE), and analyze architectural factors (circuit complexity or depth, convergence) alongside per-emotion robustness. QNNs generally outperform the CNN-Base under pitch shift, temporal shift, and speed variation (up to 22% lower CE/RCE at severe temporal shift), while the CNN-Base remains more resilient to Gaussian noise. Among quantum circuits, QNN-Basic achieves the best overall robustness on AVFAD, and QNN-Random performs strongest on TESS. Emotion-wise, fear is most robust (80-90% accuracy under severe corruptions), neutral can collapse under strong Gaussian noise (5.5% accuracy), and happy is most vulnerable to pitch, temporal, and speed distortions. QNNs also converge up to six times faster than the CNN-Base. To our knowledge, this is a systematic study of QNN robustness for speech under common non-adversarial acoustic corruptions, indicating that shallow entangling quantum front-ends can improve noise resilience while sensitivity to additive noise remains a challenge. --- ### WebCoderBench: Benchmarking Web Application Generation with Comprehensive and Interpretable Evaluation Metrics Authors: Chenxu Liu, Yingjie Fu, Wei Yang, Ying Zhang, Tao Xie Published: 2026-01-05 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2601.02430v3 Web applications (web apps) have become a key arena for large language models (LLMs) to demonstrate their code generation capabilities and commercial potential. However, building a benchmark for LLM-generated web apps remains challenging due to the need for real-world user requirements, generalizable evaluation metrics without relying on ground-truth implementations or test cases, and interpretable evaluation results. To address these challenges, we introduce WebCoderBench, the first real-world-collected, generalizable, and interpretable benchmark for web app generation. WebCoderBench comprises 1,572 real user requirements, covering diverse modalities and expression styles that reflect realistic user intentions. WebCoderBench provides 24 fine-grained evaluation metrics across 9 perspectives, combining rule-based and LLM-as-a-judge paradigm for fully automated, objective, and general evaluation. Moreover, WebCoderBench adopts human-preference-aligned weights over metrics to yield interpretable overall scores. Experiments across 12 representative LLMs and 2 LLM-based agents show that there exists no dominant model across all evaluation metrics, offering an opportunity for LLM developers to optimize their models in a targeted manner for a more powerful version. --- ### Heterogeneous Low-Bandwidth Pre-Training of LLMs Authors: Yazan Obeidi, Amir Sarfi, Joel Lidin, Paul Janson, Eugene Belilovsky Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02360v1 Pre-training large language models (LLMs) increasingly requires distributed compute, yet bandwidth constraints make it difficult to scale beyond well-provisioned datacenters-especially when model parallelism forces frequent, large inter-device communications. We study whether SparseLoCo, a low-communication data parallel method based on infrequent synchronization and sparse pseudo-gradient exchange, can be combined with low-bandwidth pipeline model parallelism via activation and activation-gradient compression. We introduce a heterogeneous distributed training framework where some participants host full replicas on high-bandwidth interconnects, while resource-limited participants are grouped to jointly instantiate a replica using pipeline parallelism with subspace-projected inter-stage communication. To make the recently introduced subspace pipeline compression compatible with SparseLoCo, we study a number of adaptations. Across large-scale language modeling experiments (178M-1B parameters) on standard pretraining corpora, we find that activation compression composes with SparseLoCo at modest cost, while selective (heterogeneous) compression consistently improves the loss-communication tradeoff relative to compressing all replicas-especially at aggressive compression ratios. These results suggest a practical path to incorporating low-bandwidth model parallelism and heterogeneous participants into LLM pre-training. --- ### ExposeAnyone: Personalized Audio-to-Expression Diffusion Models Are Robust Zero-Shot Face Forgery Detectors Authors: Kaede Shiohara, Toshihiko Yamasaki, Vladislav Golyanik Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02359v1 Detecting unknown deepfake manipulations remains one of the most challenging problems in face forgery detection. Current state-of-the-art approaches fail to generalize to unseen manipulations, as they primarily rely on supervised training with existing deepfakes or pseudo-fakes, which leads to overfitting to specific forgery patterns. In contrast, self-supervised methods offer greater potential for generalization, but existing work struggles to learn discriminative representations only from self-supervision. In this paper, we propose ExposeAnyone, a fully self-supervised approach based on a diffusion model that generates expression sequences from audio. The key idea is, once the model is personalized to specific subjects using reference sets, it can compute the identity distances between suspected videos and personalized subjects via diffusion reconstruction errors, enabling person-of-interest face forgery detection. Extensive experiments demonstrate that 1) our method outperforms the previous state-of-the-art method by 4.22 percentage points in the average AUC on DF-TIMIT, DFDCP, KoDF, and IDForge datasets, 2) our model is also capable of detecting Sora2-generated videos, where the previous approaches perform poorly, and 3) our method is highly robust to corruptions such as blur and compression, highlighting the applicability in real-world face forgery detection. --- ### VINO: A Unified Visual Generator with Interleaved OmniModal Context Authors: Junyi Chen, Tong He, Zhoujie Fu, Pengfei Wan, Kun Gai, Weicai Ye Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02358v2 We present VINO, a unified visual generator that performs image and video generation and editing within a single framework. Instead of relying on task-specific models or independent modules for each modality, VINO uses a shared diffusion backbone that conditions on text, images and videos, enabling a broad range of visual creation and editing tasks under one model. Specifically, VINO couples a vision-language model (VLM) with a Multimodal Diffusion Transformer (MMDiT), where multimodal inputs are encoded as interleaved conditioning tokens, and then used to guide the diffusion process. This design supports multi-reference grounding, long-form instruction following, and coherent identity preservation across static and dynamic content, while avoiding modality-specific architectural components. To train such a unified system, we introduce a multi-stage training pipeline that progressively expands a video generation base model into a unified, multi-task generator capable of both image and video input and output. Across diverse generation and editing benchmarks, VINO demonstrates strong visual quality, faithful instruction following, improved reference and attribute preservation, and more controllable multi-identity edits. Our results highlight a practical path toward scalable unified visual generation, and the promise of interleaved, in-context computation as a foundation for general-purpose visual creation. --- ### DARC: Drum accompaniment generation with fine-grained rhythm control Authors: Trey Brosnan Published: 2026-01-05 Categories: cs.SD, cs.AI, eess.AS arXiv: https://arxiv.org/abs/2601.02357v1 In music creation, rapid prototyping is essential for exploring and refining ideas, yet existing generative tools often fall short when users require both structural control and stylistic flexibility. Prior approaches in stem-to-stem generation can condition on other musical stems but offer limited control over rhythm, and timbre-transfer methods allow users to specify specific rhythms, but cannot condition on musical context. We introduce DARC, a generative drum accompaniment model that conditions both on musical context from other stems and explicit rhythm prompts such as beatboxing or tapping tracks. Using parameter-efficient fine-tuning, we augment STAGE, a state-of-the-art drum stem generator, with fine-grained rhythm control while maintaining musical context awareness. --- ### Talk2Move: Reinforcement Learning for Text-Instructed Object-Level Geometric Transformation in Scenes Authors: Jing Tan, Zhaoyang Zhang, Yantao Shen, Jiarui Cai, Shuo Yang, Jiajun Wu, Wei Xia, Zhuowen Tu, Stefano Soatto Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02356v2 We introduce Talk2Move, a reinforcement learning (RL) based diffusion framework for text-instructed spatial transformation of objects within scenes. Spatially manipulating objects in a scene through natural language poses a challenge for multimodal generation systems. While existing text-based manipulation methods can adjust appearance or style, they struggle to perform object-level geometric transformations-such as translating, rotating, or resizing objects-due to scarce paired supervision and pixel-level optimization limits. Talk2Move employs Group Relative Policy Optimization (GRPO) to explore geometric actions through diverse rollouts generated from input images and lightweight textual variations, removing the need for costly paired data. A spatial reward guided model aligns geometric transformations with linguistic description, while off-policy step evaluation and active step sampling improve learning efficiency by focusing on informative transformation stages. Furthermore, we design object-centric spatial rewards that evaluate displacement, rotation, and scaling behaviors directly, enabling interpretable and coherent transformations. Experiments on curated benchmarks demonstrate that Talk2Move achieves precise, consistent, and semantically faithful object transformations, outperforming existing text-guided editing approaches in both spatial accuracy and scene coherence. --- ### Meta-Learning Guided Pruning for Few-Shot Plant Pathology on Edge Devices Authors: Mohammed Mudassir Uddin, Shahnawaz Alam, Mohammed Kaif Pasha, Dr Tasneem Bano Rehman, Dr Fahmina Taranum, Afroze Begum Published: 2026-01-05 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.02353v3 Farmers in remote areas need quick and reliable methods for identifying plant diseases, yet they often lack access to laboratories or high-performance computing resources. Deep learning models can detect diseases from leaf images with high accuracy, but these models are typically too large and computationally expensive to run on low-cost edge devices such as Raspberry Pi. Furthermore, collecting thousands of labeled disease images for training is both expensive and time-consuming. This paper addresses both challenges by combining neural network pruning, removing unnecessary parts of the model, with few-shot learning, which enables the model to learn from limited examples. This paper proposes Disease-Aware Channel Importance Scoring (DACIS), a method that identifies which parts of the neural network are most important for distinguishing between different plant diseases, integrated into a three-stage Prune-then-Meta-Learn-then-Prune (PMP) pipeline. Experiments on PlantVillage and PlantDoc datasets demonstrate that the proposed approach reduces model size by 78% while maintaining 92.3% of the original accuracy, with the compressed model running at 7 frames per second on a Raspberry Pi 4, making real-time field diagnosis practical for smallholder farmers. --- ### Falcon-H1R: Pushing the Reasoning Frontiers with a Hybrid Model for Efficient Test-Time Scaling Authors: Falcon LLM Team, Iheb Chaabane, Puneesh Khanna, Suhail Mohmad, Slim Frikha, Shi Hu, Abdalgader Abubaker, Reda Alami, Mikhail Lubinets, Mohamed El Amine Seddik, Hakim Hacid Published: 2026-01-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.02346v1 This work introduces Falcon-H1R, a 7B-parameter reasoning-optimized model that establishes the feasibility of achieving competitive reasoning performance with small language models (SLMs). Falcon-H1R stands out for its parameter efficiency, consistently matching or outperforming SOTA reasoning models that are $2\times$ to $7\times$ larger across a variety of reasoning-intensive benchmarks. These results underscore the importance of careful data curation and targeted training strategies (via both efficient SFT and RL scaling) in delivering significant performance gains without increasing model size. Furthermore, Falcon-H1R advances the 3D limits of reasoning efficiency by combining faster inference (through its hybrid-parallel architecture design), token efficiency, and higher accuracy. This unique blend makes Falcon-H1R-7B a practical backbone for scaling advanced reasoning systems, particularly in scenarios requiring extensive chain-of-thoughts generation and parallel test-time scaling. Leveraging the recently introduced DeepConf approach, Falcon-H1R achieves state-of-the-art test-time scaling efficiency, offering substantial improvements in both accuracy and computational cost. As a result, Falcon-H1R demonstrates that compact models, through targeted model training and architectural choices, can deliver robust and scalable reasoning performance. --- ### Joint Semantic and Rendering Enhancements in 3D Gaussian Modeling with Anisotropic Local Encoding Authors: Jingming He, Chongyi Li, Shiqi Wang, Sam Kwong Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02339v1 Recent works propose extending 3DGS with semantic feature vectors for simultaneous semantic segmentation and image rendering. However, these methods often treat the semantic and rendering branches separately, relying solely on 2D supervision while ignoring the 3D Gaussian geometry. Moreover, current adaptive strategies adapt the Gaussian set depending solely on rendering gradients, which can be insufficient in subtle or textureless regions. In this work, we propose a joint enhancement framework for 3D semantic Gaussian modeling that synergizes both semantic and rendering branches. Firstly, unlike conventional point cloud shape encoding, we introduce an anisotropic 3D Gaussian Chebyshev descriptor using the Laplace-Beltrami operator to capture fine-grained 3D shape details, thereby distinguishing objects with similar appearances and reducing reliance on potentially noisy 2D guidance. In addition, without relying solely on rendering gradient, we adaptively adjust Gaussian allocation and spherical harmonics with local semantic and shape signals, enhancing rendering efficiency through selective resource allocation. Finally, we employ a cross-scene knowledge transfer module to continuously update learned shape patterns, enabling faster convergence and robust representations without relearning shape information from scratch for each new scene. Experiments on multiple datasets demonstrate improvements in segmentation accuracy and rendering quality while maintaining high rendering frame rates. --- ### Robust Persona-Aware Toxicity Detection with Prompt Optimization and Learned Ensembling Authors: Berk Atil, Rebecca J. Passonneau, Ninareh Mehrabi Published: 2026-01-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.02337v1 Toxicity detection is inherently subjective, shaped by the diverse perspectives and social priors of different demographic groups. While ``pluralistic'' modeling as used in economics and the social sciences aims to capture perspective differences across contexts, current Large Language Model (LLM) prompting techniques have different results across different personas and base models. In this work, we conduct a systematic evaluation of persona-aware toxicity detection, showing that no single prompting method, including our proposed automated prompt optimization strategy, uniformly dominates across all model-persona pairs. To exploit complementary errors, we explore ensembling four prompting variants and propose a lightweight meta-ensemble: an SVM over the 4-bit vector of prompt predictions. Our results demonstrate that the proposed SVM ensemble consistently outperforms individual prompting methods and traditional majority-voting techniques, achieving the strongest overall performance across diverse personas. This work provides one of the first systematic comparisons of persona-conditioned prompting for toxicity detection and offers a robust method for pluralistic evaluation in subjective NLP tasks. --- ### BEDS : Bayesian Emergent Dissipative Structures : A Formal Framework for Continuous Inference Under Energy Constraints Authors: Laurent Caraffa Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02329v2 We introduce BEDS (Bayesian Emergent Dissipative Structures), a formal framework for analyzing inference systems that must maintain beliefs continuously under energy constraints. Unlike classical computational models that assume perfect memory and focus on one-shot computation, BEDS explicitly incorporates dissipation (information loss over time) as a fundamental constraint. We prove a central result linking energy, precision, and dissipation: maintaining a belief with precision $τ$ against dissipation rate $γ$ requires power $P \geq γk_{\rm B} T / 2$, with scaling $P \propto γ\cdot τ$. This establishes a fundamental thermodynamic cost for continuous inference. We define three classes of problems -- BEDS-attainable, BEDS-maintainable, and BEDS-crystallizable -- and show these are distinct from classical decidability. We propose the Gödel-Landauer-Prigogine conjecture, suggesting that closure pathologies across formal systems, computation, and thermodynamics share a common structure. --- ### Hunting for "Oddballs" with Machine Learning: Detecting Anomalous Exoplanets Using a Deep-Learned Low-Dimensional Representation of Transit Spectra with Autoencoders Authors: Alexander Roman, Emilie Panek, Roy T. Forestano, Eyup B. Unlu, Katia Matcheva, Konstantin T. Matchev Published: 2026-01-05 Categories: astro-ph.EP, astro-ph.IM, cs.LG arXiv: https://arxiv.org/abs/2601.02324v1 This study explores the application of autoencoder-based machine learning techniques for anomaly detection to identify exoplanet atmospheres with unconventional chemical signatures using a low-dimensional data representation. We use the Atmospheric Big Challenge (ABC) database, a publicly available dataset with over 100,000 simulated exoplanet spectra, to construct an anomaly detection scenario by defining CO2-rich atmospheres as anomalies and CO2-poor atmospheres as the normal class. We benchmarked four different anomaly detection strategies: Autoencoder Reconstruction Loss, One-Class Support Vector Machine (1 class-SVM), K-means Clustering, and Local Outlier Factor (LOF). Each method was evaluated in both the original spectral space and the autoencoder's latent space using Receiver Operating Characteristic (ROC) curves and Area Under the Curve (AUC) metrics. To test the performance of the different methods under realistic conditions, we introduced Gaussian noise levels ranging from 10 to 50 ppm. Our results indicate that anomaly detection is consistently more effective when performed within the latent space across all noise levels. Specifically, K-means clustering in the latent space emerged as a stable and high-performing method. We demonstrate that this anomaly detection approach is robust to noise levels up to 30 ppm (consistent with realistic space-based observations) and remains viable even at 50 ppm when leveraging latent space representations. On the other hand, the performance of the anomaly detection methods applied directly in the raw spectral space degrades significantly with increasing the level of noise. This suggests that autoencoder-driven dimensionality reduction offers a robust methodology for flagging chemically anomalous targets in large-scale surveys where exhaustive retrievals are computationally prohibitive. --- ### Environment-Adaptive Covariate Selection: Learning When to Use Spurious Correlations for Out-of-Distribution Prediction Authors: Shuozhi Zuo, Yixin Wang Published: 2026-01-05 Categories: stat.ME, cs.LG arXiv: https://arxiv.org/abs/2601.02322v2 A common approach to out-of-distribution prediction restricts models to causal or invariant covariates to avoid spurious associations that may change across environments. Despite its theoretical appeal, this strategy can underperform empirical risk minimization when only a subset of the causal parents of the outcome is observed. In such settings, non-causal covariates can serve as proxies for unobserved causal parents and improve prediction when the proxy relationship is stable, but they can hurt when shifts disrupt that relationship. Thus, the optimal covariate set can depend on the specific shift encountered. Because different shifts leave signatures in the unlabeled covariate distribution, we propose an environment-adaptive covariate selection algorithm that maps environment-level summaries to environment-specific covariate sets. These summaries may be hand-crafted or learned from multi-environment data, and prior causal knowledge can be incorporated as constraints. Across simulations and applied datasets, the proposed method improves over static causal, invariant, and other non-adaptive rules under diverse shifts. --- ### Estimating Text Temperature with Language Models Authors: Nikolay Mikhaylovskiy Published: 2026-01-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.02320v2 Autoregressive language models typically use temperature parameter at inference to shape the probability distribution and control the randomness of the text generated. After the text was generated, this parameter can be estimated using maximum likelihood approach. Following it, we propose a procedure to estimate the temperature of any text, including ones written by humans, with respect to a given language model. We evaluate the temperature estimation capability of a wide selection of small-to-medium Large Language Models (LLMs). We then use the best-performing Qwen3 14B to estimate temperatures of popular corpora, finding that while most measured temperatures are close to 1, notable exceptions include Jokes, GSM8K, and AG News (1.1), and Python code (0.9). --- ### Fusion2Print: Deep Flash-Non-Flash Fusion for Contactless Fingerprint Matching Authors: Roja Sahoo, Anoop Namboodiri Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02318v2 Contactless fingerprint recognition offers a hygienic and convenient alternative to contact-based systems, enabling rapid acquisition without latent prints, pressure artifacts, or hygiene risks. However, contactless images often show degraded ridge clarity due to illumination variation, subcutaneous skin discoloration, and specular reflections. Flash captures preserve ridge detail but introduce noise, whereas non-flash captures reduce noise but lower ridge contrast. We propose Fusion2Print (F2P), the first framework to systematically capture and fuse paired flash-non-flash contactless fingerprints. We construct a custom paired dataset, FNF Database, and perform manual flash-non-flash subtraction to isolate ridge-preserving signals. A lightweight attention-based fusion network also integrates both modalities, emphasizing informative channels and suppressing noise, and then a U-Net enhancement module produces an optimally weighted grayscale image. Finally, a deep embedding model with cross-domain compatibility, generates discriminative and robust representations in a unified embedding space compatible with both contactless and contact-based fingerprints for verification. F2P enhances ridge clarity and achieves superior recognition performance (AUC=0.999, EER=1.12%) over single-capture baselines (Verifinger, DeepPrint). --- ### DatBench: Discriminative, Faithful, and Efficient VLM Evaluations Authors: DatologyAI, :, Siddharth Joshi, Haoli Yin, Rishabh Adiga, Ricardo Monti, Aldo Carranza, Alex Fang, Alvin Deng, Amro Abbas, Brett Larsen, Cody Blakeney et al. Published: 2026-01-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.02316v2 Empirical evaluation serves as the primary compass guiding research progress in foundation models. Despite a large body of work focused on training frontier vision-language models (VLMs), approaches to their evaluation remain nascent. To guide their maturation, we propose three desiderata that evaluations should satisfy: (1) faithfulness to the modality and application, (2) discriminability between models of varying quality, and (3) efficiency in compute. Through this lens, we identify critical failure modes that violate faithfulness and discriminability, misrepresenting model capabilities: (i) multiple-choice formats reward guessing, poorly reflect downstream use cases, and saturate early as models improve; (ii) blindly solvable questions, which can be answered without images, constitute up to 70% of some evaluations; and (iii) mislabeled or ambiguous samples compromise up to 42% of examples in certain datasets. Regarding efficiency, the computational burden of evaluating frontier models has become prohibitive: by some accounts, nearly 20% of development compute is devoted to evaluation alone. Rather than discarding existing benchmarks, we curate them via transformation and filtering to maximize fidelity and discriminability. We find that converting multiple-choice questions to generative tasks reveals sharp capability drops of up to 35%. In addition, filtering blindly solvable and mislabeled samples improves discriminative power while simultaneously reducing computational cost. We release DatBench-Full, a cleaned evaluation suite of 33 datasets spanning nine VLM capabilities, and DatBench, a discriminative subset that achieves 13x average speedup (up to 50x) while closely matching the discriminative power of the original datasets. Our work outlines a path toward evaluation practices that are both rigorous and sustainable as VLMs continue to scale. --- ### Prithvi-Complimentary Adaptive Fusion Encoder (CAFE): unlocking full-potential for flood inundation mapping Authors: Saurabh Kaushik, Lalit Maurya, Beth Tellman Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02315v1 Geo-Foundation Models (GFMs), have proven effective in diverse downstream applications, including semantic segmentation, classification, and regression tasks. However, in case of flood mapping using Sen1Flood11 dataset as a downstream task, GFMs struggles to outperform the baseline U-Net, highlighting model's limitation in capturing critical local nuances. To address this, we present the Prithvi-Complementary Adaptive Fusion Encoder (CAFE), which integrate Prithvi GFM pretrained encoder with a parallel CNN residual branch enhanced by Convolutional Attention Modules (CAM). Prithvi-CAFE enables fast and efficient fine-tuning through adapters in Prithvi and performs multi-scale, multi-level fusion with CNN features, capturing critical local details while preserving long-range dependencies. We achieve state-of-the-art results on two comprehensive flood mapping datasets: Sen1Flood11 and FloodPlanet. On Sen1Flood11 test data, Prithvi-CAFE (IoU 83.41) outperforms the original Prithvi (IoU 82.50) and other major GFMs (TerraMind 82.90, DOFA 81.54, spectralGPT: 81.02). The improvement is even more pronounced on the hold-out test site, where Prithvi-CAFE achieves an IoU of 81.37 compared to the baseline U-Net (70.57) and original Prithvi (72.42). On FloodPlanet, Prithvi-CAFE also surpasses the baseline U-Net and other GFMs, achieving an IoU of 64.70 compared to U-Net (60.14), Terramind (62.33), DOFA (59.15) and Prithvi 2.0 (61.91). Our proposed simple yet effective Prithvi-CAFE demonstrates strong potential for improving segmentation tasks where multi-channel and multi-modal data provide complementary information and local details are critical. The code is released on \href{https://github.com/Sk-2103/Prithvi-CAFE}{Prithvi-CAFE Github} --- ### Project Ariadne: A Structural Causal Framework for Auditing Faithfulness in LLM Agents Authors: Sourena Khanzadeh Published: 2026-01-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.02314v1 As Large Language Model (LLM) agents are increasingly tasked with high-stakes autonomous decision-making, the transparency of their reasoning processes has become a critical safety concern. While \textit{Chain-of-Thought} (CoT) prompting allows agents to generate human-readable reasoning traces, it remains unclear whether these traces are \textbf{faithful} generative drivers of the model's output or merely \textbf{post-hoc rationalizations}. We introduce \textbf{Project Ariadne}, a novel XAI framework that utilizes Structural Causal Models (SCMs) and counterfactual logic to audit the causal integrity of agentic reasoning. Unlike existing interpretability methods that rely on surface-level textual similarity, Project Ariadne performs \textbf{hard interventions} ($do$-calculus) on intermediate reasoning nodes -- systematically inverting logic, negating premises, and reversing factual claims -- to measure the \textbf{Causal Sensitivity} ($φ$) of the terminal answer. Our empirical evaluation of state-of-the-art models reveals a persistent \textit{Faithfulness Gap}. We define and detect a widespread failure mode termed \textbf{Causal Decoupling}, where agents exhibit a violation density ($ρ$) of up to $0.77$ in factual and scientific domains. In these instances, agents arrive at identical conclusions despite contradictory internal logic, proving that their reasoning traces function as "Reasoning Theater" while decision-making is governed by latent parametric priors. Our findings suggest that current agentic architectures are inherently prone to unfaithful explanation, and we propose the Ariadne Score as a new benchmark for aligning stated logic with model action. --- ### Game of Coding: Coding Theory in the Presence of Rational Adversaries, Motivated by Decentralized Machine Learning Authors: Hanzaleh Akbari Nodehi, Viveck R. Cadambe, Mohammad Ali Maddah-Ali Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02313v1 Coding theory plays a crucial role in enabling reliable communication, storage, and computation. Classical approaches assume a worst-case adversarial model and ensure error correction and data recovery only when the number of honest nodes exceeds the number of adversarial ones by some margin. However, in some emerging decentralized applications, particularly in decentralized machine learning (DeML), participating nodes are rewarded for accepted contributions. This incentive structure naturally gives rise to rational adversaries who act strategically rather than behaving in purely malicious ways. In this paper, we first motivate the need for coding in the presence of rational adversaries, particularly in the context of outsourced computation in decentralized systems. We contrast this need with existing approaches and highlight their limitations. We then introduce the game of coding, a novel game-theoretic framework that extends coding theory to trust-minimized settings where honest nodes are not in the majority. Focusing on repetition coding, we highlight two key features of this framework: (1) the ability to achieve a non-zero probability of data recovery even when adversarial nodes are in the majority, and (2) Sybil resistance, i.e., the equilibrium remains unchanged even as the number of adversarial nodes increases. Finally, we explore scenarios in which the adversary's strategy is unknown and outline several open problems for future research. --- ### Placement Semantics for Distributed Deep Learning: A Systematic Framework for Analyzing Parallelism Strategies Authors: Deep Pankajbhai Mehta Published: 2026-01-05 Categories: cs.DC, cs.AI arXiv: https://arxiv.org/abs/2601.02311v1 Training large language models requires distributing computation across many accelerators, yet practitioners select parallelism strategies (data, tensor, pipeline, ZeRO) through trial and error because no unified systematic framework predicts their behavior. We introduce placement semantics: each strategy is specified by how it places four training states (parameters, optimizer, gradients, activations) across devices using five modes (replicated, sharded, sharded-with-gather, materialized, offloaded). From placement alone, without implementation details, we derive memory consumption and communication volume. Our predictions match published results exactly: ZeRO-3 uses 8x less memory than data parallelism at 1.5x communication cost, as reported in the original paper. We prove two conditions (gradient integrity, state consistency) are necessary and sufficient for distributed training to match single-device results, and provide composition rules for combining strategies safely. The framework unifies ZeRO Stages 1-3, Fully Sharded Data Parallel (FSDP), tensor parallelism, and pipeline parallelism as instances with different placement choices. --- ### Temporal Kolmogorov-Arnold Networks (T-KAN) for High-Frequency Limit Order Book Forecasting: Efficiency, Interpretability, and Alpha Decay Authors: Ahmad Makinde Published: 2026-01-05 Categories: cs.LG, q-fin.TR arXiv: https://arxiv.org/abs/2601.02310v2 High-Frequency trading (HFT) environments are characterised by large volumes of limit order book (LOB) data, which is notoriously noisy and non-linear. Alpha decay represents a significant challenge, with traditional models such as DeepLOB losing predictive power as the time horizon (k) increases. In this paper, using data from the FI-2010 dataset, we introduce Temporal Kolmogorov-Arnold Networks (T-KAN) to replace the fixed, linear weights of standard LSTMs with learnable B-spline activation functions. This allows the model to learn the 'shape' of market signals as opposed to just their magnitude. This resulted in a 19.1% relative improvement in the F1-score at the k = 100 horizon. The efficacy of T-KAN networks cannot be understated, producing a 132.48% return compared to the -82.76% DeepLOB drawdown under 1.0 bps transaction costs. In addition to this, the T-KAN model proves quite interpretable, with the 'dead-zones' being clearly visible in the splines. The T-KAN architecture is also uniquely optimized for low-latency FPGA implementation via High level Synthesis (HLS). The code for the experiments in this project can be found at https://github.com/AhmadMak/Temporal-Kolmogorov-Arnold-Networks-T-KAN-for-High-Frequency-Limit-Order-Book-Forecasting. --- ### 360DVO: Deep Visual Odometry for Monocular 360-Degree Camera Authors: Xiaopeng Guo, Yinzhe Xu, Huajian Huang, Sai-Kit Yeung Published: 2026-01-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.02309v2 Monocular omnidirectional visual odometry (OVO) systems leverage 360-degree cameras to overcome field-of-view limitations of perspective VO systems. However, existing methods, reliant on handcrafted features or photometric objectives, often lack robustness in challenging scenarios, such as aggressive motion and varying illumination. To address this, we present 360DVO, the first deep learning-based OVO framework. Our approach introduces a distortion-aware spherical feature extractor (DAS-Feat) that adaptively learns distortion-resistant features from 360-degree images. These sparse feature patches are then used to establish constraints for effective pose estimation within a novel omnidirectional differentiable bundle adjustment (ODBA) module. To facilitate evaluation in realistic settings, we also contribute a new real-world OVO benchmark. Extensive experiments on this benchmark and public synthetic datasets (TartanAir V2 and 360VO) demonstrate that 360DVO surpasses state-of-the-art baselines (including 360VO and OpenVSLAM), improving robustness by 50% and accuracy by 37.5%. Homepage: https://chris1004336379.github.io/360DVO-homepage --- ### Differential Privacy for Transformer Embeddings of Text with Nonparametric Variational Information Bottleneck Authors: Dina El Zein, James Henderson Published: 2026-01-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.02307v2 We propose a privacy-preserving method for sharing text data by sharing noisy versions of their transformer embeddings. It has been shown that hidden representations learned by deep models can encode sensitive information from the input, making it possible for adversaries to recover the input data with considerable accuracy. This problem is exacerbated in transformer embeddings because they consist of multiple vectors, one per token. To mitigate this risk, we propose Nonparametric Variational Differential Privacy (NVDP), which ensures both useful data sharing and strong privacy protection. We take a differential privacy (DP) approach, integrating a nonparametric variational information bottleneck (NVIB) layer into the transformer architecture to inject noise into its multivector embeddings and thereby hide information, and measuring privacy protection with Rényi Divergence (RD) and its corresponding Bayesian Differential Privacy (BDP) guarantee. Training the NVIB layer calibrates the noise level according to the utility of the downstream task. We test NVDP on the General Language Understanding Evaluation (GLUE) benchmark and show that varying the noise level gives us a useful trade-off between privacy and accuracy. With lower noise levels, our model maintains high accuracy while offering strong privacy guarantees, effectively balancing privacy and utility. --- ### Classifying several dialectal Nawatl varieties Authors: Juan-José Guzmán-Landa, Juan-Manuel Torres-Moreno, Miguel Figueroa-Saavedra, Carlos-Emiliano González-Gallardo, Graham Ranger, Martha Lorena-Avendaño-Garrido Published: 2026-01-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.02303v1 Mexico is a country with a large number of indigenous languages, among which the most widely spoken is Nawatl, with more than two million people currently speaking it (mainly in North and Central America). Despite its rich cultural heritage, which dates back to the 15th century, Nawatl is a language with few computer resources. The problem is compounded when it comes to its dialectal varieties, with approximately 30 varieties recognised, not counting the different spellings in the written forms of the language. In this research work, we addressed the problem of classifying Nawatl varieties using Machine Learning and Neural Networks. ---