knn.zip_ANN_KNN ANN SVM_SVM KNN 识别_knn svm ann_knn 分类器

function Q_all=knn(train,test,kkk) % Usage: [Cmat,C_rate]=knn(Pr,Tr,Pt,Tt,kN) % kNN- k-Nearest Neighbor Classifier % train: training feature vector K x N (prototype), the classifier % test: testing/validating feature vector Q x N %[train]=normalize(train); %[test]=normalize(test); [K,N]=size(train); Q=size(test,1); d=zeros(K,Q); train_v=train(:,1:N-1); train_l=train(:,N); test_v=test(:,1:N-1); test_l=test(:,N); clear train test % calculate L2 norm distance class=[]; for i=1:Q for j=1:K d(j,i)=sum((train_v(j,:)-test_v(i,:)).^2); end end clear train_v test_v for i=1:Q [mm,index]=sort(d(:,i)); k=train_l(index(1:kkk)); if (sum(k)>=1.0) cl=1; else cl=-1; end class=[class;cl]; end QQ=0;QP=0;QN=0; TP=0;TN=0;FP=0;FN=0; for i=1:Q if (test_l(i)==1.0) if (class(i)==1.0) TP=TP+1; else FN=FN+1; end elseif(test_l(i)==-1.0) if (class(i)==-1.0) TN=TN+1; else FP=FP+1; end end end QQ=(TP+TN)/(TP+TN+FP+FN); QP=TP/(TP+FN); QN=TN/(TN+FP); p=0;n=0; for i=1:K if train_l(i)==1 p=p+1; elseif lable_tr(i)==-1 n=n+1; end end Q_all=[p,n,TP,FN,TN,FP,QP,QN,QQ];