arXiv AI 论文日报 — 2026-08-15

arXiv AI 论文日报 — 2026-08-15

抓取时间: 08:15> 来源: arXiv.org (cs.AI / cs.LG / cs.CL / cs.CV / cs.MA)> 论文总量: 33 篇—## 🔥 今日热门### 1. Intern-S2-Preview: Scientific Agentic Foundation Model- 📂分类: 机器学习 (ML) | 🔥热度: 100/100- 🔗 https://arxiv.org/abs/2608.13505v1- 👤 Lei Bai, Jiaqi Cao, Chiyu Chen, Guanzhou Chen, Kai Chen et al.- 📅 2026-08-13-摘要: Scientific discovery increasingly requires AI systems that can reason over scientific evidence of heterogeneous modalities, interact with scientific tools and environments, and sustain progress across long task horizons. We present Intern-S2-Preview, a series of scientific agentic foundation models …### 2. MLLM-Routed Heterogeneous Ensembles for Robust Cross-Dataset Image Classification- 📂分类: 计算机视觉 (CV) | 🔥热度: 85/100- 🔗 https://arxiv.org/abs/2608.13463v1- 👤 Daniel Perkins, John Squires, Janou Milligan, Chandra Raskoti, Linda Ungerboeck- 📅 2026-08-13-摘要: Modern image classification models excel when trained on single task-specific datasets but often struggle to generalize across domains and difficulty levels. We propose ARMDIL, an Adaptive Router for Multi-Domain Image classification with LLMs. ARMDIL is an ensemble that uses a multimodal large lang…### 3. Reduced Matrix Multiplication: Input-Adaptive Matrix-Product Reduction for LLM Inference- 📂分类: 机器学习 (ML) | 🔥热度: 75/100- 🔗 https://arxiv.org/abs/2608.13426v1- 👤 Zixuan Lan, Yanhong Li, Jiawei Zhou- 📅 2026-08-13-摘要: Transformer-based language models achieve strong performance but incur substantial inference cost due to repeated high-dimensional matrix multiplications. We propose Reduced Matrix Multiplication (RMM), a training-free, input-adaptive inference method that reduces Transformer matrix products by sele…### 4. TraVEL: Trajectory-Guided Video Embedding Learning for Driving-Video Retrieval- 📂分类: 计算机视觉 (CV) | 🔥热度: 75/100- 🔗 https://arxiv.org/abs/2608.13495v1- 👤 Yi-Chung Chen, Philip Jacobson, Tom Lampo, Yiren Lu, Jin Yao et al.- 📅 2026-08-13-摘要: Efficiently retrieving relevant clips from large-scale driving logs is essential for data curation, model development, and safety analysis. Structured and rule-based retrieval systems can explicitly target driving events, but typically require expert-defined rules, auxiliary data, and multi-stage pe…### 5. AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design- 📂分类: 计算机视觉 (CV) | 🔥热度: 55/100- 🔗 https://arxiv.org/abs/2608.13560v1- 👤 Yaxin Luo, Haobin Jiang, Jialv Zou, Xu Huang, Wenhao Yan et al.- 📅 2026-08-13-摘要: Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a model-harness system. While an ideal harness system should align with human design priors and accumulate reusable experience through empiric…## 📂 分类浏览### 机器学习 (ML) (7篇)-🔥 Vero: Can AI Agents Build Formally Verified Software Repositories?- https://arxiv.org/abs/2608.13522v1 - Zhe Ye, Hantao Lou, Yuechun Sun, Peiyang Song, Zhengxu Yan et al. | 2026-08-13-📄 The data geometry of masking diffusion: Certified-optimal schedules via unmasking growth complexity- https://arxiv.org/abs/2608.13520v1 - Martin J. Wainwright | 2026-08-13-📄 Synthetic Persona Pretraining: Alignment from Token Zero- https://arxiv.org/abs/2608.13482v1 - Julian Minder, Viktor Moskvoretskii, Raghav Singhal, Difan Jiao, Andy Arditi et al. | 2026-08-13-📄 Concept Drift Detection and Adaptive Retraining of Malware Classification Models- https://arxiv.org/abs/2608.13465v1 - Christofer Washington Berruz Chungata, Martin Jurecek, Katerina Potika, William B. Andreopoulos, Mark Stamp | 2026-08-13-🔥 Intern-S2-Preview: Scientific Agentic Foundation Model- https://arxiv.org/abs/2608.13505v1 - Lei Bai, Jiaqi Cao, Chiyu Chen, Guanzhou Chen, Kai Chen et al. | 2026-08-13-🔥 Reduced Matrix Multiplication: Input-Adaptive Matrix-Product Reduction for LLM Inference- https://arxiv.org/abs/2608.13426v1 - Zixuan Lan, Yanhong Li, Jiawei Zhou | 2026-08-13-📄 Intervention-Aware Clinical World Model for Post-Op Outcome Forecasting in Cardiology- https://arxiv.org/abs/2608.13518v1 - Yunsung Chung, Yingshuo Liu, Abboud F. Hassan, Han Feng, Mary M. Maleckar et al. | 2026-08-13### 人工智能 (AI) (4篇)-🔥 OmniScientist: An Omni-Modal Omni-Discipline AI Scientist- https://arxiv.org/abs/2608.13558v1 - Bobo Li, Hao Fei, Tianjie Ju, Mong-Li Lee, Wynne Hsu | 2026-08-13-📄 QuoteBench: How Matched Scores Can Hide Command-Path Failures- https://arxiv.org/abs/2608.13547v1 - Shangao Li, Yao Zhang, Volker Tresp, Yuanyuan Yang | 2026-08-13-📄 AlayaWorld: Interactive Long-Horizon World Modeling - Full Technical Report (v1.1)- https://arxiv.org/abs/2608.13492v1 - AlayaWorld Team, Kaipeng Zhang, Chuanhao Li, Yifan Zhan, Yongtao Ge et al. | 2026-08-13-🔥 MARC v1: An Open-Source Multi-Agent Framework for Clinical AI Reasoning and Coordination- https://arxiv.org/abs/2608.13476v1 - Saisha Shetty, Satvik Tripathi, Austin Lin, Colin Zhao, Theodore Kim et al. | 2026-08-13### 计算语言学 (NLP) (8篇)-📄 LittleLearner: Language Models Under Pedagogically Controlled Knowledge Exposure- https://arxiv.org/abs/2608.13545v1 - Fanfei Li, Jana Zeller, Manuel Prada-Corral, Thaddäus Wiedemer, Prasanna Mayilvahanan et al. | 2026-08-13-🔥 DFM Mimir v1: An Open HRM Delivering Frontier Performance at 1B Parameters Using Only Permissible Post-Training Data- https://arxiv.org/abs/2608.13517v1 - Peter Schneider-Kamp, Jacob Nielsen, Gianluca Barmina, Kenneth Enevoldsen, Lukas Galke Poech | 2026-08-13-🔥 Toward a Gricean Retreat: Probing LLMs for Knowledge Boundaries and Referent Specificity- https://arxiv.org/abs/2608.13484v1 - Dananjay Srinivas, Saksham Khatwani, Maria Pacheco | 2026-08-13-📄 SAEVerbalizer: Generating Explanations for Sparse Autoencoder Features via Representation Verbalization- https://arxiv.org/abs/2608.13538v1 - Weihan Meng, Hongzhu Guo, Yi Jing, Dewen Liu, Zijun Yao et al. | 2026-08-13-📄 Measuring Task-Agnostic Training Data Influence Across Language Model Pretraining- https://arxiv.org/abs/2608.13515v1 - Yuto Nishida, Hirokazu Kiyomaru, Yusuke Oda, Takashi Kodama, Chaoran Liu et al. | 2026-08-13-📄 Are You Sure You’re Sure? On the Impact of Instruction Tuning on Confidence and Lexical Diversity- https://arxiv.org/abs/2608.13430v1 - Irina Proskurina, Mayank Kumar, Oyindolapo O. Komolafe | 2026-08-13-📄 Motor, Cognitive, or Corpus? What Survives Cross-Lingual Transfer in Speech-Based Parkinsons Disease Detection- https://arxiv.org/abs/2608.13425v1 - Serli Kopar, Sam Gijsen, Abner Hernandez, Paula Andrea Perez-Toro, Kerstin Ritter | 2026-08-13-🔥 CROP: Task Relevance via Counterfactuals for Selective On-Policy Distillation- https://arxiv.org/abs/2608.13387v1 - Enhan Li, Junhao He, Hongyang Du | 2026-08-13### 计算机视觉 (CV) (12篇)-🔥 AutoDesign: Meta-Harness Optimization for Long-Horizon Agentic Design- https://arxiv.org/abs/2608.13560v1 - Yaxin Luo, Haobin Jiang, Jialv Zou, Xu Huang, Wenhao Yan et al. | 2026-08-13-🔥 MLLM-Routed Heterogeneous Ensembles for Robust Cross-Dataset Image Classification- https://arxiv.org/abs/2608.13463v1 - Daniel Perkins, John Squires, Janou Milligan, Chandra Raskoti, Linda Ungerboeck | 2026-08-13-📄 V-RAE: Rethinking Video Latent Spaces for Generation- https://arxiv.org/abs/2608.13556v1 - Minghui Guo, Shengqiong Wu, Hao Fei | 2026-08-13-🔥 PlayWorld: Benchmarking World Models with Agent Players over Long-Horizon Objectives- https://arxiv.org/abs/2608.13552v1 - Kaixin Ding, Xi Chen, Minghong Cai, Zhiyuan Xu, Yiyang Wang et al. | 2026-08-13-📄 Alaya-EVOKE: From Linear-Scaling Supervision to Endless World- https://arxiv.org/abs/2608.13546v1 - Yuanyang Yin, Gongxuan Wang, Yifan Zhan, Chuanhao Li, Kaipeng Zhang et al. | 2026-08-13-🔥 SCULPT: Subtractive Composition for 3D Part Generation- https://arxiv.org/abs/2608.13541v1 - Sikuang Li, Chen Yang, Jiemin Fang, Jiazhong Cen, Yuhe Wei et al. | 2026-08-13-🔥 TabSOM: A tabular-to-image encoding method based on self-organizing maps- https://arxiv.org/abs/2608.13513v1 - David Chushig-Muzo, María Ángeles Rodríguez de Cara, Eva Milara, Francisco J. Lara-Abelenda, Luis Zhinin-Vera et al. | 2026-08-13-🔥 GS2^{2}2CI: Robust Gaussian Splatting For Snapshot Compressive Imaging via Large Vision Model Priors- https://arxiv.org/abs/2608.13502v1 - Yanming Yang, Chenxi Song, Ping Wang, Xin Yuan, Chi Zhang | 2026-08-13-🔥 TraVEL: Trajectory-Guided Video Embedding Learning for Driving-Video Retrieval- https://arxiv.org/abs/2608.13495v1 - Yi-Chung Chen, Philip Jacobson, Tom Lampo, Yiren Lu, Jin Yao et al. | 2026-08-13-📄 DreamX-Phi 1.0: Action-Conditioned Video World Model for Robotic Manipulation- https://arxiv.org/abs/2608.13489v1 - DreamX Team, Rui Chen, Xiangxiang Chu, Geng Li, Jifan Li et al. | 2026-08-13-🔥 MapRoute++: Surrogate-Guided Semantic Routing for Visual Concept Unlearning- https://arxiv.org/abs/2608.13478v1 - Ashok Urlana, L. D. M. S. Sai Teja, Vivek Hruday Kavuri, Ponnurangam Kumaraguru | 2026-08-13-🔥 SNM-VFI: Symmetric Nonlinear Motion-Guided Generative Video Frame Interpolation- https://arxiv.org/abs/2608.13460v1 - Jisoo Jeong, Hong Cai, Jamie Menjay Lin, Hanno Ackermann, Hyeonjun Sim et al. | 2026-08-13### 多智能体系统 (1篇)-🔥 AaLLM: An End-to-End Analog Circuit Design Framework from Topology Generation to Sizing Using Large Language Models- https://arxiv.org/abs/2608.13472v1 - Mohammed Ayman Habib, Rylan Hart, Morteza Fayazi | 2026-08-13### 机器人学 (1篇)-🔥 HumanTracker: Towards Comprehensive and Human-Aligned Motion Tracking Benchmark- https://arxiv.org/abs/2608.13555v1 - Dairu Liu, Zekun Qi, Jiayu Zeng, Ruixi Yu, Yu Guan et al. | 2026-08-13—日报由 arXiv Daily Bot 自动生成 | 2026-08-15 08:15