arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

University of Science and Technology of China(中国科学技术大学)

2026-01-30 至 2026-01-30 共收录 7
2601.21938 2026-01-30 cs.CV

BookNet: Book Image Rectification via Cross-Page Attention Network

BookNet:通过跨页注意力网络实现书本图像校正

Shaokai Liu, Hao Feng, Bozhi Luan, Min Hou, Jiajun Deng, Wengang Zhou

机构 * Hefei University of Technology(合肥工业大学) University of Science and Technology of China(中国科学技术大学)

AI总结 BookNet通过跨页注意力机制实现双页书本图像校正,引入Book3D和Book100数据集,优于现有方法。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21797 2026-01-30 cs.CL

Enhancing Conversational Agents via Task-Oriented Adversarial Memory Adaptation

通过任务导向的对抗性记忆适应增强对话代理

Yimin Deng, Yuqing Fu, Derong Xu, Yejing Wang, Wei Ni, Jingtong Gao, Xiaopeng Li, Chengxu Liu, Xiao Han, Guoshuai Zhao, Xiangyu Zhao, Li Zhu, Xueming Qian

机构 * Xi’an Jiaotong University(西安交通大学) City University of Hong Kong(香港城市大学) University of Science and Technology of China(中国科学技术大学) Zhejiang University(浙江大学) Zhejiang University of Technology(浙江工业大学)

AI总结 本文提出对抗性记忆适应机制,通过模拟任务执行提升对话代理对长对话的处理能力。

Comments 11 pages, 4 figures

详情

展开后加载摘要…

URL PDF HTML 收藏
2508.19325 2026-01-30 cs.CV

PRISM: A Framework Harnessing Unsupervised Visual Representations and Textual Prompts for Explainable MACE Survival Prediction from Cardiac Cine MRI

PRISM:一种结合无监督视觉表示和文本提示的框架,用于从心脏 cine MRI 预测可解释的 MACE 生存率

Haoyang Su, Jin-Yi Xiang, Shaohao Rui, Yifan Gao, Xingyu Chen, Tingxuan Yin, Shaoting Zhang, Xiaosong Wang, Lian-Ming Wu

机构 * Fudan University(复旦大学) Shanghai Innovation Institute(上海创新研究院) Shanghai Artificial Intelligence Laboratory(上海人工智能实验室) Shanghai Jiao Tong University(上海交通大学) University of Science and Technology of China(中国科学技术大学)

AI总结 PRISM通过结合无监督视觉表示和文本提示,实现了从心脏 cine MRI 预测可解释的 MACE 生存率。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21709 2026-01-30 cs.CL

Why Attention Patterns Exist: A Unifying Temporal Perspective Analysis

为何存在注意力模式:一种统一的时间视角分析

Qingyue Yang, Jie Wang, Xing Li, Yinqi Bai, Xialiang Tong, Huiling Zhen, Jianye Hao, Mingxuan Yuan, Bin Li

机构 * MoE Key Laboratory of Brain-inspired Intelligent Perception and Cognition, University of Science and Technology of China(脑启发智能感知与认知教育部重点实验室,中国科学技术大学) Huawei Technologies Co., Ltd.(华为技术有限公司) College of Intelligence and Computing, Tianjin University(智能与计算学院,天津大学)

AI总结 本文提出TAPPA框架,从时间连续性角度分析注意力模式,揭示其可预测性与查询自相似性关系,并通过KV缓存压缩和剪枝任务验证其有效性。

Comments ICLR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21381 2026-01-30 cs.LG

DA-SPS: A Dual-stage Network based on Singular Spectrum Analysis, Patching-strategy and Spearman-correlation for Multivariate Time-series Prediction

DA-SPS:基于奇异谱分析、补丁策略和斯皮尔曼相关性的双阶段网络用于多变量时间序列预测

Tianhao Zhang, Shusen Ma, Yu Kang, Yun-Bo Zhao

机构 * Department of Automation, University of Science and Technology of China(自动化系,中国科学技术大学) Institute of Advanced Technology, University of Science and Technology of China(先进技术研究院,中国科学技术大学) Institute of Artificial Intelligence, Hefei Comprehensive National Science Center(人工智能研究院,合肥综合性国家科学中心)

AI总结 DA-SPS通过双阶段网络结合奇异谱分析、补丁策略和斯皮尔曼相关性,提升多变量时间序列预测的准确性与鲁棒性。

Comments 12 pages, 7 figures, 6 tables, submitted to IEEE Transactions on Emerging Topics in Computational Intelligence

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.21320 2026-01-30 cs.CV cs.LG

Optimal Transport-Induced Samples against Out-of-Distribution Overconfidence

最优传输诱导的对抗性样本对抗分布外过度自信

Keke Tang, Ziyong Du, Xiaofei Wang, Weilong Peng, Peican Zhu, Zhihong Tian

机构 * Guangzhou University(广州大学) SmartMore Corporation(SmartMore公司) University of Science and Technology of China(中国科学技术大学) Northwestern Polytechnical University(西北工业大学) Guangdong Key Laboratory of Industrial Control System Security(广东工业控制系统安全重点实验室)

AI总结 本文提出通过最优传输诱导的边界生成分布外样本,以缓解模型在不确定区域的过度自信预测问题。

Comments Accepted by ICLR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2512.11229 2026-01-30 cs.CV cs.SD

REST: Diffusion-based Real-time End-to-end Streaming Talking Head Generation via ID-Context Caching and Asynchronous Streaming Distillation

REST:基于扩散模型的实时端到端流式说话人生成方法:通过ID上下文缓存和异步流式蒸馏

Haotian Wang, Yuzhe Weng, Jun Du, Haoran Xu, Xiaoyan Wu, Shan He, Bing Yin, Cong Liu, Qingfeng Liu

机构 * University of Science and Technology of China(中国科学技术大学)

AI总结 REST通过ID-Context缓存和异步流式蒸馏,实现高效实时端到端流式说话人生成。

Comments 27 pages, 10 figures

详情

展开后加载摘要…

URL PDF HTML 收藏