美赛常见参考代码;特征降维，特征融合，相关分析等多元数据分析的典型相关分析代码.zip

function [ SolVecs ] = CCA_zq ( X , M ,ClassNum ) %X ---训练样本特征向量（列向量），把每个样本的两组特征向量，按照特征向量的维数由小到大首尾相连。 %M ---存储两组特征向量维数的数组(列向量) %ClassNum ---类别数 %SolVecs ---求得的近似解向量 %Author ---陆凤娟、周强（本程序是在陆凤娟师姐程序基础上直接截取添加所得） 南京理工大学603教研室 moxibingdao@qq.com % %以下为添加改动部分 pcafeature=X(1:M(1),:); TrainImage1=pcafeature; otherfeature=X(1+M(1):M(1)+M(2),:); TrainImage2=otherfeature; cn=ClassNum; train_NUM=size(X,2)/ClassNum; N=(train_NUM)*cn; vec_dim1=M(1); vec_dim2=M(2); % %以上为添加改动部分 disp('计算Sxx...'); %求Swx Sxx = zeros(vec_dim1,vec_dim1); I = eye(vec_dim1); for i =1:cn within_xx = zeros(vec_dim1,vec_dim1); % temp = TrainImage1(:,(i-1)*train_NUM + 1:i*train_NUM ); for j=1:train_NUM % x = TrainImage1(:,(i-1)*train_NUM + j) - mean2(TrainImage1(:,(i-1)*train_NUM + 1:i*train_NUM )); x = TrainImage1(:,(i-1)*train_NUM + j) - mean(TrainImage1,2);%mean(temp,2); within_xx = within_xx + x*x'; end; Sxx = Sxx + within_xx; end; Sxx = Sxx ; disp('计算Syy...'); %求Syy Syy = zeros(vec_dim2,vec_dim2); I = eye(vec_dim2); for i =1:cn within_yy = zeros(vec_dim2,vec_dim2); % temp = TrainImage2(:,(i-1)*train_NUM + 1:i*train_NUM ); for j=1:train_NUM y = TrainImage2(:,(i-1)*train_NUM + j) - mean(TrainImage2,2);% mean(temp,2); within_yy = within_yy + y*y'; end; Syy = Syy + within_yy; end; Syy = Syy; disp('计算Lxy...'); %求Lxy Lxy = zeros(vec_dim1,vec_dim2); for i=1:size(TrainImage1,2) x = TrainImage1(:,i) - mean(TrainImage1,2); y = TrainImage2(:,i) - mean(TrainImage2,2); % xclass = ceil(i/train_NUM);%这幅图是第几类 % yclass = ceil(i/train_NUM); % x = TrainImage1(:,i) - mean(TrainImage1(:,(xclass-1)*train_NUM+1:xclass*train_NUM),2); % y = TrainImage2(:,i) - mean(TrainImage2(:,(yclass-1)*train_NUM+1:yclass*train_NUM),2); Lxy = Lxy + x*y'; end; Lxy = Lxy; Lyx = Lxy'; disp('Lxy的秩：'); dxy = rank(Lxy) %%%CVs A = inv(Sxx)*Lxy*inv(Syy)*Lyx; A=A+0.0000001*(eye(size(A)));;%防止奇异 [Vec1,lamda1] = eig(A); % if rank(A)==size(A,1) % [Vec1,lamda1] = eig(A); % else % A=A+eye(size(A,1))*0.000001; % end; B = inv(Syy)*Lyx*inv(Sxx)*Lxy; %B=B+0.1*eye(size(B));; [Vec2,lamda2] = eig(B); % if rank(B)==size(B,1) % [Vec2,lamda2] = eig(B); % else % B=B+eye(size(B,1))*0.000001; % end; % [Vec2,lamda2] = eig(B); %sort the vectors lamda1 = diag(lamda1); [lamda1 ,pos1]=sort(lamda1,'descend'); lamda2 = diag(lamda2); [lamda2 ,pos2]=sort(lamda2,'descend'); %解两个方程归一化 for i=1:length(pos1) temp1= Vec1(:,pos1(i)); CVx(:,i) = temp1; % CVx(:,i) = temp/sqrt(temp'*Sxx*temp); temp2 = Vec2(:,pos2(i)); CVy(:,i) = temp2; % CVy(:,i) = temp/sqrt(temp'*Syy*temp); end; alfa=CVx; beta=CVy; SolVecs=[ alfa beta ];