The paper introduces SLIM-init, a monocular visual‑inertial initialization method that avoids explicit 3D reconstruction by using 2D line features and vanishing points. It provides robust rotation constraints through line‑derived vanishing points and improves translation estimation with a line epipolar residual and a line‑normal projection residual. Experiments on public benchmarks and custom degenerate‑motion sequences show that SLIM-init achieves higher accuracy and robustness than existing initialization techniques.
By Junwan Choi, Woongrae Jo, Dong-Uk Seo, Jinwoo Jeon, Hyun Myung
The paper investigates Japanese past‑tense verb inflection in neural morphological generation systems, focusing on a rare irregular subtype where stems end in /e/ and require gemination before the suffix. Despite overall high accuracy (>97%), this <1% subclass accounts for 30–43% of residual errors, amplifying its impact by 34–48 times its frequency. Ablation experiments show that removing this specific subtype yields larger accuracy gains than removing all irregular verbs, highlighting the role of low‑frequency patterns and orthographic processes in error concentration.
By Wen Zhang
VGGT-CAD is a geometry‑aware framework that reconstructs parametric CAD 3D models from single and multi‑view images. It incorporates pretrained 3D geometric priors by encoding camera parameters as condition tokens and jointly modeling them with image tokens. The method introduces a variable‑view cross‑view context aggregation module and a training‑free geometry‑aware view selection strategy, and decodes the learned representation into CAD command sequences using a non‑autoregressive decoder. Additionally, VideoCAD, a large‑scale multi‑view video benchmark derived from existing CAD data, is presented to evaluate the approach.
By Chunan Yu, Tianrun Chen, Fu Shen, Cheng Chen, Lanyun Zhu, Yang Yang
SafeStyle is a training‑free framework that injects calibrated style residuals into frozen diffusion models for reference‑guided stylization. It estimates style‑supported and content‑associated subspaces from small calibration sets, then transports purified style evidence across adaptive spatial granularity while limiting its influence with a residual‑norm budget. Experiments on texture‑ and geometry‑dominant styles show high DINO style similarity (0.432–0.474) with minimal semantic leakage (0.8%).
By Zhangping Yang, Min Li, Song Yan, Rong Gao, Xinliang Bi, Guanye Xiong, Yujie He
S3VD is a new video deraining framework that leverages semantic guidance and spatio‑temporal scanning to improve performance over existing State Space Models such as Mamba. It introduces a Multi‑Scale Semantic Fusion module that uses DINOv2 priors to preserve 2D spatial semantics, and a Spatio‑Temporal Scanning Fusion module that incorporates a Decoupled‑Gating Mamba layer to better model intra‑ and inter‑frame correlations. Experiments on video deraining benchmarks show that S3VD achieves state‑of‑the‑art results, improving PSNR by an average of 0.84 dB over Mamba‑based baselines.
By Kui Jiang, Yiang Chen, Yan Luo, Zhaocheng Yu, Junjun Jiang, Xianming Liu
Large Language Model agents are being explored for chip design, but most methods work directly at RTL. This study compares Direct RTL Design, Agent-based HLS Design, Post-Compiler HLS Refinement, and Post-HLS RTL Refinement, and proposes a combined workflow called Agent-based HLS with RTL Refinement (AHRR). Using FPGAs for evaluation, AHRR achieves a 2.6× geometric‑mean speedup over Direct RTL Design across an 11‑task benchmark suite, demonstrating that higher‑level abstractions and subsequent RTL refinement can improve chip design efficiency.
By Zijian Ding, Yang Zou, Yizhou Sun, Jason Cong
The paper introduces PSEE, a method for progressive sensor event expansion that generates pseudo action segments from point-supervised labels in wearable sensor streams. By leveraging semantic activations, sensor-specific transition evidence, and adaptive temporal ownership, PSEE produces high‑quality pseudo boundaries that can train standard temporal action localization detectors without altering their inference. Experiments on four inertial‑sensing benchmarks show that PSEE outperforms adapted point‑supervised baselines, works with various detectors, and remains robust to different point sampling strategies.
By Jiaxi Yin, Ge Wang, Han Ding, Fei Wang
SpecOpt is a new molecular design task that optimizes the binding specificity of existing drugs by making constrained structural modifications. The method uses an agentic framework that docks a compound against its intended target and known off‑targets, compares residue‑aware atom‑protein contacts, and feeds the differential interactions to a large language model to propose changes. On a benchmark of 915 compounds, SpecOpt increased the target‑off‑target binding gap for 84.8% of cases while preserving drug‑like properties and structural similarity.
By Thao Nguyen, Heng Ji
CogGym is a scalable, unified framework that standardizes diverse cognitive experiments into a task‑agnostic Experiment Markup Language (EML) for systematic comparison of human and AI behavior. The initial release curates 258 experiments from 100 papers focused on human commonsense reasoning and evaluates 50 large language models, revealing a scaling trend where larger models better reproduce human judgments but still lag far behind human split‑half reliability. The framework aims to continually incorporate new cognitive science experiments to track where model behavior aligns with or diverges from human cognition as models evolve.
By Lance Ying, Jinzhou Wu, Yingshan Susan Wang, Shivam Aarya, Luca M. Schulze Buschoff, Harry Chen, Katherine M. Collins, Andrea de Varda, Shuhao Fu, Sean Dae Houlihan, Akshay K. Jagadish, Guangyuan Jiang, Samuel Kiegeland, Tetsu Kurumisawa, Rongzhi Liu, Ryan Liu, Ningshan Ma, Kathryn McGregor, Younes Strittmatter, Polina Tsvilodub, Jacob Hoover Vigly, Sarah Wu, Enjie Xu, Yiling Yun, Kelsey Allen, Tyler Brooke-Wilson, Brian Christian, Evelina Fedorenko, Michael C. Frank, Michael Franke, Tao Gao, Samuel J. Gershman, Robert D. Hawkins, Jennifer Hu, Julian Jara-Ettinger, Max Kleiman-Weiner, Sydney Levine, Tal Linzen, Hongjing Lu, Timothy O'Donnell, Desmond C. Ong, Steven T. Piantadosi, Rebecca Saxe, Eric Schulz, Tianmin Shu, Felix A. Sosa, Ilia Sucholutsky, Tan Zhi-Xuan, Tomer Ullman, Fei Xu, Ilker Yildirim, Jian-Qiao Zhu, Thomas L. Griffiths, Tobias Gerstenberg, Kevin Smith, Joshua B. Tenenbaum
OmniVChat defines a native audio‑visual dialogue task where models receive raw audio and video from a user and produce text responses, eliminating the need for separate text queries or speech recognition. To address data scarcity and evaluation challenges, the authors introduce OmniVChat‑Studio, a multi‑agent engine that synthesizes single‑ and multi‑turn dialogues, and OmniVChat‑Bench, a benchmark assessing models across five dialogue ability categories. They also propose OmniVChat‑RL, a reinforcement‑learning reward that balances reply correctness, efficiency, and style, and demonstrate that training Qwen3‑Omni‑Instruct with this reward on synthesized data improves performance on both synthetic and human‑recorded benchmarks.
By Haolin He, Yunfei Chu, Qi Chen, Wen Huang, Yuan Feng, Muzhi Zhu, Zheqi Dai, Haoning Xu, Dongchao Yang, Chunyat Wu, Zining Liang, Zhengxi Liu, Xiquan Li, Xie Chen, Xize Cheng, Qize Yang, Jin Xu, Qiuqiang Kong
The paper introduces a Chinese competitive debating dataset and benchmark, comprising 148 fully adjudicated matches with 2,698 stages and 20,542 exchange units, all annotated by 120 professional judges using a shared rubric. It offers manually verified transcripts, segmentation, and preserved scores, votes, and rationales, enabling evaluation of large language models on winner-tendency prediction, stage-score prediction, and best-debater prediction. Zero‑shot experiments show the best models achieve 66.2% winner‑prediction accuracy, a 0.250 Pearson correlation for stage scores, and 56.8% best‑debater accuracy.
By Zongrui Yang, Haoyuan Li, Zhongsheng Wang, Zhirui Zeng, Pengqian Han, Yi Zhou, Yuting Wang, Jiamou Liu
Samsone is a family of small audio language models (SALMs) designed for on‑device inference, with the flagship Samsone‑134M setting a new state‑of‑the‑art for its size class across multiple benchmarks. The paper also presents Samsone‑99M and Samsone‑356M to study scaling laws, showing that these compact models achieve performance competitive with much larger counterparts. The authors train the models on publicly available data and release training code, weights, mobile‑optimized checkpoints, and an open‑source Android app for real‑time inference.
By Piotr Masztalski, Micha{\l} K. Grzeszczyk, Olaf Sikorski
The paper introduces a new multi-draft speculative sampling algorithm that uses Poisson processes to improve inference efficiency and output provenance for large language models. It achieves strong sampling efficiency while allowing an unbiased watermark to be embedded without reducing speculative acceptance. The method relies on an exact list‑coupling‑without‑communication scheme, giving a drafter‑invariant property that benefits both sampling and watermarking, and the authors experimentally confirm its effectiveness.
By Yanxiao Liu, Sicheng Wan, Zhan Gao, Deniz G\"und\"uz
ProTracer is a training‑free framework that uses Vision‑Language Models (VLMs) together with proprioceptive signals to analyze robot manipulation failures. It performs binary failure detection, categorization, explanation generation, and introduces failure onset localization—identifying the earliest moment a robot deviates from a valid trajectory leading to failure. The method leverages proprioceptive dynamics to pinpoint informative action boundaries and converts robot‑state signals into natural‑language descriptions for joint multimodal reasoning, achieving strong performance on both conventional failure diagnosis and the new failure onset localization task.
By Chang Dong, Mehdi Hosseinzadeh, King Hang Wong, Lingqiao Liu, Francois Fraysse, Feras Dayoub, Minh Hoai Nguyen
The paper proposes a Bayesian inverse problem formulation for unsupervised anomaly detection, aiming to infer the most probable corruption causing each observation. This approach yields a probabilistic anomaly score based on the energy of inferred corruption parameters and provides a unified framework that recovers several existing methods as special cases. Experiments demonstrate improved object-class AUROC on the MVTec AD dataset and strong detection performance on a brain MRI benchmark, with additional estimates of pathology intensity, bias, and geometry.
By James Myles, Matthew Baugh, Johanna P. M\"uller, Bernhard Kainz, Yingzhen Li
Particle Competition and Cooperation for Robust Graph Convolutional Network Learning Under Label Noise proposes PCC+GCN, a hybrid framework that refines labels using Particle Competition and Cooperation before training a GCN. PCC identifies suspicious nodes via particle domination dynamics and decides whether to keep, remove, or reassign their labels, optionally augmenting the graph with k‑nearest‑neighbor edges. Evaluated on ten NoisyGL datasets under various noise types, PCC+GCN achieved the highest average accuracy and rank, outperforming baseline GCN by 1.67 percentage points and proving computationally efficient, especially under instance‑dependent noise.
By Fabricio Breve
The paper introduces a compositional continual learning benchmark for world models in robot manipulation, designed to isolate knowledge reuse from learning speed and capacity. Tasks are curated to combine previously seen action and perception components, allowing analysis of how different modalities affect reuse. Experiments show that modular world models better balance reuse and forgetting than conventional methods, yet none fully solve the challenge, highlighting the need for models explicitly built to reuse knowledge without forgetting.
By Haoyu Zhou, Joe Watson, Anson Lei, Ingmar Posner
TALON is a Temporally Aware Longitudinal framework for radiology report generation that adaptively integrates variable‑length patient histories. Its Dual‑Channel Temporal Fusion Module compares each prior examination with the current one through similarity and change channels, using channel‑specific attention and a prior‑specific gate to weigh relevance and suppress redundancy. Experiments on MIMIC‑CXR demonstrate that TALON surpasses state‑of‑the‑art methods on clinical efficacy and graph‑based metrics, with performance improving as more prior examinations are available.
By Nien-Tsyr Sun, Min-Chen Chen, Hui Nien Hung, Vincent S. Tseng
VISPATH is a visual‑intent‑guided path reasoning framework designed for multimodal knowledge graph question answering (MM‑KGQA). It first identifies a reliable starting entity by fusing multimodal grounding with graph‑structural cues, then iteratively discovers and refines reasoning paths using hop‑specific multimodal intent and a reasoning‑chain pruning step. The framework is evaluated on the newly introduced VISPATH‑Bench, which tests two‑to‑four‑hop reasoning, and demonstrates consistent improvements over strong baselines, even surpassing GPT‑5.4 when using GPT‑4o as the backbone.
By Jinke Wu, Zhengpin Li, Mengzhe Jia, Yang Li, Wentao Zhang
PhysioBench is a unified benchmark for physiological signal question answering that consolidates annotations from 22 public datasets into 61.4 million question‑answer pairs spanning 30 tasks. Each pair is linked to a specific signal segment and traceable to its source annotation. The benchmark evaluates 21 models—including large language models, vision‑language models, time‑series language models, and physiological signal foundation models—under three settings, revealing that no model consistently excels across all modalities and tasks, and that performance is sensitive to question wording.
By Mengxuan Li, Junfa Chen, Jinze Xia, Yundan Chen, Lixin Fan, Ke Liu, Keyue Shi, Haishuai Wang