The paper introduces ENFR, a method for estimating the absolute evidence ζ of a data‑generating model in Bayesian inference. It learns Bayes factors relative to a flow‑matched reference and then recovers the absolute evidence, improving over existing flow‑based baselines across multiple applications. The authors also provide a joint architecture for estimating evidences of multiple models and demonstrate its performance and scalability on diverse scientific datasets.
By Kai Lehman, Jameson Dong, Carolina Cuesta-Lazaro, Niall Jeffrey, Benjamin D. Wandelt, Francisco Villaescusa-Navarro, Sven Krippendorf, Jochen Weller
The paper introduces RoboHarness, a simple robotic harness that equips large language models (LLMs) with a visual‑geometric control panel for direct manipulation of embodied tasks. By enabling LLM agents to understand and invoke actions without complex vision‑language‑action (VLA) training, RoboHarness allows models such as Qwen3.8‑Flash‑Next to outperform state‑of‑the‑art VLA models on the BEHAVIOR Challenge 2025 tasks. The authors provide an open‑source implementation of the system.
By Bince Qu, Wei Chen, Bo Zhang
The paper investigates self‑supervised pretraining for Handwritten Text Recognition (HTR) and finds that pixel‑reconstruction methods outperform contrastive methods, a contrast to natural‑image classification. The authors argue that discriminative information for HTR is concentrated in high‑variance pixel directions, and they test six SSL methods across six handwriting benchmarks in five languages, showing that pixel‑grounded approaches achieve the lowest character error rates and are more label‑efficient. When combined with a pretrained LLM decoder, a frozen pixel‑grounded encoder rivals fully fine‑tuned supervised baselines, underscoring the importance of aligning representations with high‑variance pixel subspaces.
By Carlos Garrido-Munoz, Jorge Calvo-Zaragoza
The paper investigates position bias in large language models (LLMs) across a wide range of model sizes—from billions to trillions of parameters—and across five languages. It finds that while small-scale models and non‑English languages still exhibit noticeable position bias, larger models and the use of Chain‑of‑Thought prompting substantially reduce this bias. Additionally, the study shows that emerging thinking models with advanced reasoning capabilities are almost immune to position bias, indicating a promising but resource‑intensive path to mitigating this vulnerability.
By Yining Li, Huayang Li, Lemao Liu, Taro Watanabe
The paper introduces a framework that aligns multimodal embeddings with language by using captioning functions on interpretable auxiliary data, a soft contrastive loss, and a rank-based concept retrieval task. Applied to Earth observation embeddings, the method improves species distribution modelling, achieving state‑of‑the‑art performance on most metrics and better concept retrieval than prior vision‑language models. The aligned concepts enable recovery of species‑habitat associations and allow domain experts to interpret model predictions in natural language.
By Gabriel\.e Tij\=unaityt\.e, Thijs L. van der Plas, Mohamed-Bachir Belaid, Rob Knapen, Maciej J. Soja, Ioannis N. Athanasiadis
UniField++ is a two‑agent framework that enhances MRI images across multiple field‑strength transitions, such as 64 mT‑to‑3 T and 3 T‑to‑7 T. It combines a VLM‑based planner that predicts risk regions and a flow‑matching enhancer that performs loop reasoning to refine high‑risk areas, guided by radiologist‑annotated hallucination‑prone regions. The method achieves significant improvements in PSNR, SSIM, and whole‑brain segmentation performance compared to existing approaches.
By Yiyang Lin, Chenhui Wang, Tianliang Yao, Shuaibo Li, Zijun Sun, Yiming Deng, Yixuan Yuan
ANCHOR is a new benchmark for audio‑visually grounded chain‑of‑thought reasoning, containing 2,319 videos and 8,272 question‑answer pairs with structured reasoning traces, spatial bounding boxes, temporal intervals, and audio cues. It evaluates not only answer correctness but also hallucination, logical divergence, and conciseness, providing direct spatial and temporal grounding measurements on a focused subset. The benchmark reveals a substantial alignment gap among contemporary multimodal large language models, showing that similar answer accuracy can mask differences in evidence‑supported explanations.
By Joel Julin, Souraja Kundu, Liza Dahiya, George Z. Wei, Liuyue Xie, Jingru Yi, Xu Zhang, Hao Yang, Laszlo A. Jeni
PersonaCUABench is a benchmark that evaluates computer‑use agents (CUAs) on 74 tasks across 11 domains, varying only the persona of the requester. Each of 44 personas rewrites the instruction in their own words and answers the CUA’s questions, while the CUA sees only the dialogue. The study finds that task completion drops by 36% when personas are introduced, and user satisfaction falls from 84.1% to 62.7% for fully completed requests.
By Haoming Li, Defu Cao, Xu Hu, Yingtao Luo, Yiyang Wang, Xianghe Xu, Jiashu He, Wanting Jiang, Yifan Zhang, Yunhan Zhao, Siyang Wu, Chen Yang, Haixin Wang, Zheyuan Deng, Taoran Li, Jingwen Zhang, Zhejian Peng, Shushu Liang, Yue Zhao, Bingzhe Li, Feng Chen, Yifan Wang, Jintao Huang, Yijun Wang, Yuexing Hao, Xiaomin Li
The paper presents a data‑augmentation framework that uses conditional generative adversarial networks and Monte Carlo dropout to synthesize structural response signals with quantified uncertainty. The synthetic signals are rigorously compared to experimental data from a BAE Systems Hawk T1A aircraft, showing accurate capture of dominant dynamics and realistic statistical properties. The study evaluates whether the augmented data can improve supervised damage classification and demonstrates the framework’s potential as a digital twin for safety‑critical aerospace structural health monitoring.
By Alvaro Peiro, Marios Impraimakis
The paper introduces extensions to the Feature Selection Layer, an embedded feature weighting mechanism for neural networks, to improve interpretability in multi‑class classification. Experiments show that the best extension matches existing methods in identifying globally relevant features while outperforming them in capturing label‑specific relevance. The approach leverages the neural network’s internal dynamics, eliminating the need for post‑processing to reveal how individual features influence specific label predictions.
By Jo\~ao Pedro Silveira e Silva, M\'arcio Dorn, Bruno Iochins Grisci
The paper introduces a stochastic variational framework using Malliavin calculus to derive non‑parametric optimality conditions for time‑varying stochastic optimization problems. It proposes the Stochastic Path Follower (SPF), a deep‑learning algorithm that remains scalable regardless of parameter‑space dimensionality. Experiments on least‑squares recovery and logistic regression under distribution drift demonstrate that SPF outperforms existing learning‑based and gradient‑based methods in both performance and scalability.
By Mohsen Amidzadeh, Lauri Viitasaari, Mario Di Francesco
The paper "Rethinking GRPO for Large Reasoning Model Safety Alignment" identifies two failure modes in Group Relative Policy Optimization—advantage collapse and unsafe diversification—that weaken safety learning in large reasoning models. It proposes the Beacon framework, which includes Safety Signal Recovery (SSR) and Safety-Utility Decoupled Alignment (SUDA), to restore effective safety signals and mitigate over-refusal. Experiments on Qwen3 and DeepSeek-R1-Distill models across eight benchmarks demonstrate that Beacon improves safety performance by up to 32.83 percentage points while keeping over-refusal low.
By Jiawei Chen, Yi Zhang, Jiawei Du, Zhonghou Lv, Tairan Chen, Junyuan Shang, Jinchang Hou, Xiao Yang, Yu Tian, Chenfu Bao, Chao Yu, Zhaoxia Yin
LongForge introduces a systematic way to evaluate and generate long‑context instruction‑tuning data by constructing evidence‑assertion reasoning graphs for each sample. The analyzer measures four dimensions—breadth, depth, correctness, and completeness—making the data auditable and revealing that breadth does not increase with document length while depth remains low. The synthesizer uses these metrics to control the training distribution, and two variants, LongForge‑1 and LongForge‑N, outperform the best baseline by 2.2–4.7 points across six benchmarks, with an additional 1.16‑point gain when explicitly increasing breadth and depth.
By Ziyang Chen, Junlong Jia, Xing Wu, Tinghao Yu, Songlin Hu
PanoBench is a new camera‑centric benchmark for panoptic occupancy prediction that unifies 3D geometry, semantics, and instance identities across synthetic and real‑world driving scenes. It combines three datasets—PanoBench‑CarlaOcc, PanoBench‑nuScenes, and PanoBench‑KITTI360—through canonical interfaces and unified evaluation protocols, providing multi‑resolution annotations with persistent instance identities. The benchmark introduces a novel reconstruction pipeline for real‑world data, a unified valid‑mask protocol, and analyses of observation fidelity, geometric completeness, spatial continuity, and temporal consistency, while also benchmarking methods on semantic, panoptic, and 4D occupancy forecasting.
whyItMatters":"PanoBench addresses the scarcity of instance‑aware annotations in panoptic occupancy research by offering comprehensive, high‑quality annotations and evaluation protocols that enable more robust development and comparison of autonomous driving perception methods."
By Yi Feng, Zizhan Guo, Junwu E, Yu Ma, Mengtan Zhang, Jin Wu, Wei Ye, Hanli Wang, Qijun Chen, Danwei Wang, Rui Fan
OmniTrace is a framework that attributes each sentence generated by an omni‑modal language model to specific source units—text chunks, images, or audio/video segments—by interpreting late‑layer attention as a mass distribution over these units, without requiring additional forward passes or per‑modality tuning. The method demonstrates that a single generation pass can recover 77–96 % of the effect of traditional leave‑one‑out attribution, which would otherwise need up to 41 extra passes. OmniTrace’s source selections are causally important for responses and align closely with human‑marked evidence, achieving higher AUROC scores than existing baselines on both document and video tasks.
By Qianqi Yan, Yichen Guo, Ching-Chen Kuo, Shan Jiang, Hang Yin, Yang Zhao, Xin Eric Wang
The paper introduces a novel SO(3)-equivariant convolutional neural network designed for data on the sphere, specifically using the Hierarchical Equal Area isoLatitude Pixelization (HEALPix). By employing steerable filters, parallel transport, and spin-weighted features, the architecture guarantees exact equivariance and avoids the need for the network to learn spherical symmetry from data. Benchmarks show the model achieves percent-level SO(3)-invariance error and outperforms existing methods in dense gradient regression, cosmic microwave background foreground removal, and denoising tasks.
By Joshua Williamson, Niall Jeffrey, Lorne Whiteway
GR-OPD introduces Group-Relative On-Policy Distillation, which separates feedback computation from credit assignment by aggregating token feedback into calibrated sequence-level credit. The method shows that broadcasting summed feedback to all response tokens provides an unbiased estimate of the negative gradient of the sequence-level reverse‑KL objective under ideal on‑policy assumptions. In practice, GR‑OPD discounts later feedback to mitigate distribution shift, uses same‑prompt rollout groups for calibration, and retains a controlled fraction of collective teacher pressure, leading to superior performance over vanilla OPD across multiple Qwen3 teacher‑student settings and mathematical reasoning benchmarks.
By Yushu Wang, Renkai Yang, Jingcao Xu, Zili Geng, Haihong Tang, Yaohui Jin, Hao He
DLite is a history‑free speculative decoding framework designed for agentic AI workloads. It bypasses the need for a historical KV cache by reusing contextual hidden states from the target model and performing local speculative predictions, supplemented by a lightweight sequential head to maintain causal dependencies. Experiments demonstrate that DLite improves acceptance length by 2–6× and outperforms the state‑of‑the‑art DSpark by up to 70%.
By Jiaming Xu, Hanzhen Wang, Tianlang Zhao, Jiewen Xiao, Chengze Yuan, Jiayi Pan, Yongkang Zhou, Jiancai Ye, Yu Wang, Guohao Dai
NestedVLA proposes a new approach for consolidating skills in Vision‑Language‑Action models by learning skills independently and then generating task‑specific parameters via a nested hypernetwork, rather than aggregating parameters. The method keeps a frozen base VLA and task‑specific compressed VLM residuals, enabling flexible, interference‑mitigated skill reuse. Experiments on MetaWorld, ManiSkill, and CALVIN show that NestedVLA outperforms the best model‑merging baseline by 41 pp on average and approaches per‑task fine‑tuning performance.
By Guodong Du, Wanyu Lin
The paper introduces an Adaptive‑Horizon Graph Neural Operator (AH‑GNO) that learns to predict bed elevation changes directly on unstructured meshes by integrating area‑weighted graph learning, historical state selection, and forecast horizon estimation. Evaluated on Telemac2D/GAIA benchmarks and a field‑scale Waiho River case, the AH‑GNO accurately reproduces key morphodynamic patterns while drastically reducing bed‑update computation time. Deployment on CPU‑GPU platforms achieves speedups of 150–360× over 4‑core GAIA CPUs and 105–186× over 20‑core CPUs.
By Ying Cao, Yifan Yang, Junqiang Xia, Julia C. Mullarney, M\'ario J. Franca