资源简介
一种矩阵补全的MATLAB实现方法
LowRank-MatrixCompletion
代码片段和文件信息
function A = completion(M)
%% create M_shadow matrix
%if the entry is known set 1
%if the entry is unknown set 0
A = M;
[mn] = size(A);
M_shadow = ones(mn);
for i = 1:m
for j = 1:n
if(isnan(M(ij)))
M_shadow(ij) = 0;
%use laplance distribution to initial the unknow entries
%Acturally this initial do not effect too much
A(ij) = 0;
end
end
end
%% Similar to gradient descent
%the reason why 1:15 is because the calculation accuracy in the matlab
%is decimal point 15
for step_size = 1:15
A = completionhelperRough(A M_shadow step_size);
A = completionhelperAccurate(A M_shadow step_size);
属性 大小 日期 时间 名称
----------- --------- ---------- ----- ----
目录 0 2016-05-15 07:56 Low-Rank-Matrix-Completion-master\
文件 280 2016-05-15 07:56 Low-Rank-Matrix-Completion-master\README.md
文件 811 2016-05-15 07:56 Low-Rank-Matrix-Completion-master\completion.m
文件 2560 2016-05-15 07:56 Low-Rank-Matrix-Completion-master\completion
文件 3910 2016-05-15 07:56 Low-Rank-Matrix-Completion-master\completion
评论
共有 条评论