arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

高校专区

University of Southern California(南加州大学)

2026-03-30 至 2026-03-30 共收录 5
2603.26078 2026-03-30 cs.CV cs.AI

When Identities Collapse: A Stress-Test Benchmark for Multi-Subject Personalization

当身份崩溃时:多主体个性化的一个压力测试基准

Zhihan Chen, Yuhuan Zhao, Yijie Zhu, Xinyu Yao

机构 * University of California, Los Angeles(加州大学洛杉矶分校) University of Southern California(南加州大学) DeerLab LLC(DeerLab有限责任公司) Carnegie Mellon University(卡内基梅隆大学)

AI总结 本文提出一个压力测试基准,揭示多主体生成模型在复杂场景下的身份崩溃问题,并引入新的评估指标SCR,强调未来生成架构需显式物理解耦。

Comments 10 pages, 7 figures, accepted by CVPR 2026 Workshop P13N

详情

展开后加载摘要…

URL PDF HTML 收藏
2603.14688 2026-03-30 cs.LG cs.AI cs.SE

AgentTrace: Causal Graph Tracing for Root Cause Analysis in Deployed Multi-Agent Systems

AgentTrace: 多智能体系统部署中的因果图追踪用于根本原因分析

Zhaohui Geoffrey Wang

机构 * USC Viterbi School of Engineering(南加州大学维特比工程学院)

AI总结 AgentTrace通过构建因果图追踪多智能体系统部署中的根本原因,实现高精度和低延迟的故障诊断,优于启发式和LLM基线方法。

Comments 11 pages, 1 figure, 19 tables. Published at ICLR 2026 Workshop on Agents in the Wild. Camera-ready version with revised layout and framework overview figure

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.22025 2026-03-30 cs.CV

Olbedo: An Albedo and Shading Aerial Dataset for Large-Scale Outdoor Environments

Olbedo:一个用于大规模户外环境的反光与阴影数据集

Shuang Song, Debao Huang, Deyan Deng, Haolin Xiong, Yang Tang, Yajie Zhao, Rongjun Qin

机构 * The Ohio State University(俄亥俄州立大学) University of Southern California(南加州大学) Institute for Creative Technologies(创意技术研究所)

AI总结 Olbedo数据集通过多视角一致的反光和阴影地图等标注,使基于合成数据训练的扩散模型在真实户外场景中实现更优的内在图像分解性能,支持3D资产光照一致重映射和城市数字孪生分析。

Comments CVPR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2508.11805 2026-03-30 eess.SY cs.NE cs.RO cs.SY

Control of a commercially available vehicle by a tetraplegic human using a brain-computer interface

通过脑机接口控制商用汽车的瘫痪人类

Xinyun Zou, Jorge Gamez, Meghna Menon, Phillip Ring, Chadwick Boulay, Likhith Chitneni, Jackson Brennecke, Shana R. Melby, Gracy Kureel, Kelsie Pejsa, Emily R. Rosario, Ausaf A. Bari, Aniruddh Ravindran, Tyson Aflalo, Spencer S. Kellis, Dimitar Filev, Florian Solzbacher, Richard A. Andersen

机构 * California Institute of Technology(加州理工学院) Ford Motor Company(福特汽车公司) Blackrock Neurotech(黑石神经技术公司) Casa Colina Hospital and Centers for Healthcare(卡萨科利纳医院与医疗中心) University of California Los Angeles(加州大学洛杉矶分校) University of Southern California(南加州大学) The University of Utah(犹他大学)

AI总结 研究通过脑机接口实现瘫痪患者在模拟和真实环境中的车辆操控,证明其与正常人相当的操控速度和精度,展示了BCI在驾驶控制中的可行性与创新性。

Comments 50 pages, 7 figures, 1 table. 27 supplementary pages, 9 supplementary figures, 13 supplementary tables, 9 supplementary movies available as ancillary files

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.10971 2026-03-30 cs.CV

ERMoE: Eigen-Reparameterized Mixture-of-Experts for Stable Routing and Interpretable Specialization

ERMoE:基于特征的混合专家架构用于稳定路由和可解释的专业化

Anzhe Cheng, Shukai Duan, Shixuan Li, Chenzhong Yin, Mingxi Cheng, Heng Ping, Tamoghna Chattopadhyay, Sophia I Thomopoulos, Shahin Nazarian, Paul Thompson, Paul Bogdan

机构 * University of Southern California(南加州大学)

AI总结 ERMoE通过重新参数化专家在学习的正交特征基上,改进路由稳定性与专家专业化可解释性,无需显式平衡损失,提升ImageNet和跨模态检索任务性能。

Comments Accepted in CVPR2026 Main Track

详情

展开后加载摘要…

URL PDF HTML 收藏