arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

Harbin Institute of Technology(哈尔滨工业大学)

2026-01-28 至 2026-01-28 共收录 7
2601.19433 2026-01-28 cs.CV

RoamScene3D: Immersive Text-to-3D Scene Generation via Adaptive Object-aware Roaming

RoamScene3D: 通过自适应物体感知漫游实现沉浸式文本到3D场景生成

Jisheng Chu, Wenrui Li, Rui Zhao, Wangmeng Zuo, Shifeng Chen, Xiaopeng Fan

机构 * Faculty of Computing, Harbin Institute of Technology(计算机学院,哈尔滨工业大学) Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences(深圳先进技术研究所,中国科学院) Nanyang Technological University(南洋理工大学) Peng Cheng Laboratory(鹏城实验室) Harbin Institute of Technology Suzhou Research Institute(哈尔滨工业大学苏州研究院)

AI总结 RoamScene3D通过自适应物体感知漫游实现沉浸式文本到3D场景生成,结合语义推理和几何约束提升场景一致性与逼真度。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.19236 2026-01-28 cs.CV cs.MM

VC-Bench: Pioneering the Video Connecting Benchmark with a Dataset and Evaluation Metrics

VC-Bench:开创视频连接基准的Dataset与评估指标

Zhiyu Yin, Zhipeng Liu, Kehai Chen, Lemao Liu, Jin Liu, Hong-Dong Li, Yang Xiang, Min Zhang

机构 * School of Computer Science and Technology, Harbin Institute of Technology, Shenzhen(计算机科学与技术学院,哈尔滨工业大学(深圳)) Peng Cheng Laboratory(鹏城实验室) School of Computer Science and Engineering, Central South University(计算机科学与工程学院,中南大学)

AI总结 VC-Bench提出一个用于视频连接任务的新型基准,包含多样化视频数据和综合评估指标,评估现有视频生成模型并揭示其在连贯性和流畅性上的不足。

详情

展开后加载摘要…

URL PDF HTML 收藏
2601.18500 2026-01-28 cs.LG

Nearly Optimal Bayesian Inference for Structural Missingness

近似最优的结构缺失性贝叶斯推断

Chen Liang, Donghua Yang, Yutong Zhao, Tianle Zhang, Shenghang Zhou, Zhiyu Liang, Hengtong Zhang, Hongzhi Wang, Ziqi Li, Xiyang Zhang, Zheng Liang, Yifei Li

机构 * Harbin Institute of Technology, Harbin, China(哈尔滨工业大学)

AI总结 本文提出了一种近似最优的贝叶斯推断框架,通过解耦模型内缺失值后验学习与标签预测,实现了在结构缺失性问题上的高效且准确的预测。

详情

展开后加载摘要…

URL PDF HTML 收藏
2511.12725 2026-01-28 cs.LG

Convolutional Model Trees

卷积模型树

William Ward Armstrong, Hongyi Li, Jun Xu

机构 * University of Alberta, Edmonton, Canada(阿尔伯塔大学) Harbin Institute of Technology, Shenzhen, China(哈尔滨工业大学)

AI总结 本文提出了一种基于卷积的模型树森林方法,通过下采样、超平面确定和卷积处理来提高图像函数拟合的准确性和平滑性。

Comments 11 pages. 2 figures. This article was extensively revised. Drawings were added. Co-authors were added responsible for cited experimental results and their description: Hongyi Li and Jun Xu. Attention is on distilling a deep net into a model tree with convolutions done on hyperplane and leaf-function coefficients. Distortions of images are treated by similar changes to coefficient locations

详情

展开后加载摘要…

URL PDF HTML 收藏
2508.10530 2026-01-28 cs.AI cs.CL

Is On-Policy Data always the Best Choice for Direct Preference Optimization-based LM Alignment?

基于直接偏好优化的LM对齐中,策略数据是否总是最佳选择?

Zetian Sun, Dongfang Li, Xuhui Chen, Baotian Hu, Min Zhang

机构 * Harbin Institute of Technology, Shenzhen(哈尔滨工业大学(深圳))

AI总结 本文提出对齐阶段假设,通过理论和实证分析,揭示了静态与策略偏好数据在LM对齐中的效果差异,并提出算法识别对齐阶段边界。

Comments Accepted by ICLR-2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2502.17498 2026-01-28 cs.LG cs.AI cs.CL

Improving Value-based Process Verifier via Structural Prior Injection

通过结构先验注入改进基于价值的过程验证器

Zetian Sun, Dongfang Li, Baotian Hu, Jun Yu, Min Zhang

机构 * Harbin Institute of Technology (Shenzhen)(哈尔滨工业大学(深圳))

AI总结 通过注入结构先验改进基于价值的过程验证器,提升其性能并减少误差。

Comments This version is deprecated. Please refer to our new version: arXiv:2508.10539

详情

展开后加载摘要…

URL PDF HTML 收藏
2502.10495 2026-01-28 cs.CR cs.AI cs.CV cs.LG

SWA-LDM: Toward Stealthy Watermarks for Latent Diffusion Models

SWA-LDM:迈向潜在扩散模型的隐蔽水印

Zhonghao Yang, Linye Lyu, Xuanhang Chang, Daojing He, YU LI

机构 * Software Engineering Institute, East China Normal University(华东师范大学软件工程学院) School of Computer Science and Technology, Harbin Institute of Technology (Shen Zhen)(哈尔滨工业大学(深圳)计算机科学与技术学院) College of Integrated Circuits, Zhejiang University(浙江大学集成电路学院)

AI总结 SWA-LDM通过动态随机化潜在噪声中的水印,提升潜在扩散模型的隐蔽性,实现更安全的水印部署。

Comments 12 pages, 5 figures

详情

展开后加载摘要…

URL PDF HTML 收藏