OPT OpenIR  > 光谱成像技术研究室
Matrix completion by Truncated Nuclear Norm Regularization
Zhang, Debing; Hu, Yao; Ye, Jieping; Li, Xuelong; He, Xiaofei
2012
会议名称2012 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2012
会议录名称2012 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2012
页码2192-2199
会议日期June 16, 2012 - June 21, 2012
会议地点Providence, RI, United states
出版地United States
出版者IEEE Computer Society, 2001 L Street N.W., Suite 700, Washington, DC 20036-4928, United States
产权排序3
摘要Estimating missing values in visual data is a challenging problem in computer vision, which can be considered as a low rank matrix approximation problem. Most of the recent studies use the nuclear norm as a convex relaxation of the rank operator. However, by minimizing the nuclear norm, all the singular values are simultaneously minimized, and thus the rank can not be well approximated in practice. In this paper, we propose a novel matrix completion algorithm based on the Truncated Nuclear Norm Regularization (TNNR)by only minimizing the smallest N-r singular values, where N is the number of singular values and r is the rank of the matrix. In this way, the rank of the matrix can be better approximated than the nuclear norm. We further develop an efficient iterative procedure to solve the optimization problem by using the alternating direction method of multipliers and the accelerated proximal gradient line search method. Experimental results in a wide range of applications demonstrate the effectiveness of our proposed approach.
作者部门光学影像分析与学习中心
收录类别CPCI(ISTP) ; EI
ISBN号9781467312264
语种英语
ISSN号10636919
文献类型会议论文
条目标识符http://ir.opt.ac.cn/handle/181661/20536
专题光谱成像技术研究室
推荐引用方式
GB/T 7714
Zhang, Debing,Hu, Yao,Ye, Jieping,et al. Matrix completion by Truncated Nuclear Norm Regularization[C]. United States:IEEE Computer Society, 2001 L Street N.W., Suite 700, Washington, DC 20036-4928, United States,2012:2192-2199.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
Matrix completion by(373KB) 限制开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[Zhang, Debing]的文章
[Hu, Yao]的文章
[Ye, Jieping]的文章
百度学术
百度学术中相似的文章
[Zhang, Debing]的文章
[Hu, Yao]的文章
[Ye, Jieping]的文章
必应学术
必应学术中相似的文章
[Zhang, Debing]的文章
[Hu, Yao]的文章
[Ye, Jieping]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。