www.pudn.com > RobustSF.zip > test.m, change:2014-09-01,size:1607b

```
downsample=3;
path=fullfile('E:\算法代码\实验AR - 统计时间-ourmethod',sprintf('Data_Mixed_downsample%d',downsample),num2str(1));
%     path=fullfile('E:\算法代码\实验AR - 统计时间-ourmethod',sprintf('Data_Scraf_downsample%d',downsample),num2str(i));
%     path=fullfile('E:\算法代码\实验AR - 统计时间-ourmethod',sprintf('Data_sumglass_downsample%d',downsample),num2str(i));

nclass=length(unique(tr_label));
%% Step1.KSVD训练字典，是一类一类的训练，然后把每一类的字典拼排在一起
singleclass=5;
D_ini=TrainDictionary(X1,tr_label,nclass,singleclass);

%% Step2. 初始化X
alpha=0.01;
beta=10;
gamma=1;
tol=1e-4;
max_iter=30;
X_ini=compute_representation(X1, D_ini, alpha, tol, max_iter);

%% Step3. 初始化W
W_ini=H*X_ini'*pinv(X_ini*X_ini'+gamma*eye(size(X_ini,1)));

%% Step4. 生成H
m=length(tr_label);
H=zeros(nclass,m);
for il=1:m
H(tr_label(il),il)=1;
end

%% Step5. 生成L
L=compute_laplase(X1);

%% Step6. 开始学习字典和低秩表示
tol=1-6;
max_iter=500;
[D,~] = DictionaryLearning(X1, X_ini, H, W_ini, D_ini, L, alpha, beta, gamma, tol, max_iter);

%% Step6. 计算测试样本的表示
[Z] = compute_representation([X1,X2], D, alpha, tol, max_iter);
Z1=Z(:,1:length(tr_label));
Z2=Z(:,lenth(tr_label)+1:lenth(tr_label)+lenth(ts_label));

%% 依据学习得到的W进行预测
[W] = compute_classifier_parametrs( Z1,tr_label);
acc  = predic_labels( Z2,ts_label,W);
fprintf('LLR without Dictionary: %f\n',acc);

```