www.pudn.com > RobustSF.zip > testparameter.m, change:2013-11-13,size:1078b


 
downsample=5; 
nround=6; 
SF_S_acc=zeros(nround,1); 
SF_W_acc=zeros(nround,1); 
SF_acc=zeros(nround,1); 
lambdaa=[0.014,0.016,0.018,0.02,0.022,0.024]; 
RobustLatLRRZ_acc=zeros(nround,1); 
RobustLatLRRLX_acc =zeros(nround,1); 
for i=1:nround 
%    path=fullfile('E:\算法代码\实验AR',sprintf('Data_sumglass_downsample%d',downsample),num2str(i)); 
%    load(fullfile(path,'ts_label.mat')); 
%    load(fullfile(path,'tr_label.mat')); 
%    load(fullfile(path,'X1.mat')); 
%    load(fullfile(path,'X2.mat')); 
 
   %合并训练样本和测试样本,然后用RPCA进行去噪,其中去噪的参数lambda对实验结果影响较大,这里取0.007 
   X=[X1 X2]; 
  [U_hat, Sigma_hat, V_hat, ~, ~, ~] = inexact_alm_rpca(X,lambdaa(i), 1e-7, 1000); %,0.007 0.028lambdaa(round)%YaleB 0.032 
 
%% RobustLatLRR Z and LX 
%   [RobustLatLRRZ_acc(i),RobustLatLRRLX_acc(i) ] = RobustLatLRR(X1,X2,U_hat, V_hat,tr_label,ts_label ); 
 
%% Our method 
lambda=1e-6; 
[SF_S_acc(i),SF_W_acc(i)] = HGHnew( U_hat,Sigma_hat,V_hat,tr_label,ts_label,lambda); 
[SF_acc(i)] = HGnew( U_hat,Sigma_hat,V_hat,tr_label,ts_label,lambda); 
end