登录    注册    忘记密码

详细信息

基于受限等距性质的矩阵低秩稀疏逼近误差界    

ERROR BOUNDS FOR SPARSE AND LOW-RANK MATRIX APPROXIMATION BASED ON RESTRICTED ISOMETRIC PROPERTIES

文献类型:期刊文献

中文题名:基于受限等距性质的矩阵低秩稀疏逼近误差界

英文题名:ERROR BOUNDS FOR SPARSE AND LOW-RANK MATRIX APPROXIMATION BASED ON RESTRICTED ISOMETRIC PROPERTIES

作者:刘子胜[1,2];李继成[1];白建超[1]

第一作者:刘子胜

机构:[1]西安交通大学数学与统计学院;[2]河南财经政法大学统计学院

第一机构:西安交通大学数学与统计学院,西安710049

年份:2018

卷号:40

期号:2

起止页码:146-159

中文期刊名:高等学校计算数学学报

外文期刊名:Numerical Mathematics A Journal of Chinese Universities

收录:CSTPCD;;北大核心:【北大核心2017】;CSCD:【CSCD_E2017_2018】;

基金:国家自然科学基金(11671318);中央高校基本科研业务费学科交叉重点项目(xkjc2014008)

语种:中文

中文关键词:数据矩阵;稀疏逼近;误差界;性质;等距;高维空间;子空间;矩阵A

外文关键词:sparse matrix;low-rank matrix;restricted isometry property;errorestimation.

摘要:1引言 假设D∈Rm×n为实际观测到的高维数据矩阵,则从高维空间中估计一低维子空间的问题,称为矩阵低秩逼近,即估计一低秩矩阵A,使得D与A∈R^m×n之间的误差E=D—A最小化,该问题表示如下。
In this paper, we consider the sparse and low-rank matrices approximation problem, which is regarded as a separable convex optimization problem subjected to linearly equality constraint. For such problem, we focus on estimating the error bounds of sparse matrix recovery. Based on the restricted isometric properties (RIP), a sufficient condition is given for exact reconstruction of sparse matrix in the ideal case. For noisy measurements, the robustness of sparse matrix restoration is analyzed and the upper bound of approximation error is also provided. Numerical simulations on solving a practical application example show that our RIP-Bound is correct.

参考文献:

正在载入数据...

版权所有©河南财经政法大学 重庆维普资讯有限公司 渝B2-20050021-8 
渝公网安备 50019002500408号 违法和不良信息举报中心