e_gaussians.m

来自「支持向量机工具箱」· M 代码 · 共 53 行

%addpath G:\Matlab_EXP\stprtool
%^addpath G:\Matlab_EXP\stprtool\data
figure('name','特征提取_高斯方法');
% Generate data
distrib.Mean = [[5;4] [4;5]]; % mean vectors
distrib.Cov(:,:,1) = [1 0.9; 0.9 1]; % 1st covariance
distrib.Cov(:,:,2) = [1 0.9; 0.9 1]; % 2nd covariance
distrib.Prior = [0.5 0.5]; % Gaussian weights
data = gmmsamp(distrib,250); % sample data
subplot(3,3,1);title('原始分布');
ppatterns(data);

lda_model = lda(data,1); % train LDA
lda_rec = pcarec(data.X,lda_model);
subplot(3,3,2);title('线性分布');

lda_data = linproj(data,lda_model);
ppatterns(lda_data);

subplot(3,3,3);title('主成分分布');

pca_model = pca(data.X,1); % train PCA
pca_rec = pcarec(data.X,pca_model);
pca_data = linproj( data,pca_model);
%figure; hold on; axis equal; % visualization
ppatterns(pca_data);

%A=lda_rec(1,:);B=lda_rec(2,:)
%h1 = plot(lda_rec(1,:),lda_rec(2,:),'r');
%h1 = plot(A,B,'r');

%pause;
%h2 = plot(pca_rec(1,:),pca_rec(2,:),'b');
;
%legend([h1 h2],'LDA direction','PCA direction');

%figure; hold on;
subplot(3,3,4);title('LDA'); 
ppatterns(lda_data);

pgauss(mlcgmm(lda_data));
subplot(3,3,5);title('PCA');

 
ppatterns(pca_data);
subplot(3,3,6);

pgauss(mlcgmm(pca_data));
subplot(3,3,7);
subplot(3,3,9);