matlab code for rotational svm

Question

alex 2019-2-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/442895-matlab-code-for-rotational-svm

编辑： alex 2019-2-9

ST7112J0_41_fea.mat

Hello. This is the code that I wrote for Rotsvm (rotational svm), but when I test it on my data set with 5 classes, the value of sensivity for each class is computed [ 0;0;1;0;0]. I do not know how to fix this problem. can any one help me plz? I attached my dataset.

clc;clear;close;
label_data=load('ST7112J0_41_fea.mat');
label_data=label_data.fea_mat;
roro=size(label_data,1);
idx1=crossvalind('kfold',roro,10);
test10 = label_data(idx1==2,:);
train90 = label_data(idx1~=2,:);
function [stats0,labelsss0,predicted_label0]=Rotsvm(train90,test10);%total input data 
train90 = [1:42;train90];
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
idx2=(crossvalind('kfold',41,6))';
train_one=train90(:,2:end);
f1=train_one(:,idx2==1);
f2=train_one(:,idx2==2);
f3=train_one(:,idx2==3);
f4=train_one(:,idx2==4);
f5=train_one(:,idx2==5);
f6=train_one(:,idx2==6);
f1=[train90(:,1) f1];
f2=[train90(:,1) f2];
f3=[train90(:,1) f3];
f4=[train90(:,1) f4];
f5=[train90(:,1) f5];
f6=[train90(:,1) f6];%ham label ham soton num asli
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
idx3=crossvalind('kfold',size(train90,1)-1,4);
ff1=f1(2:end,:);
ff2=f2(2:end,:);
ff3=f3(2:end,:);
ff4=f4(2:end,:);
ff5=f5(2:end,:);
ff6=f6(2:end,:);%just label
sub1=ff1(idx3~=1,:);
sub2=ff2(idx3~=1,:);
sub3=ff3(idx3~=1,:);
sub4=ff4(idx3~=1,:);
sub5=ff5(idx3~=1,:);
sub6=ff6(idx3~=1,:);%just label
pca_f1=pca(sub1(:,2:end),size(sub1,2)-1);
pca_f2=pca(sub2(:,2:end),size(sub2,2)-1);
pca_f3=pca(sub3(:,2:end),size(sub3,2)-1);
pca_f4=pca(sub4(:,2:end),size(sub4,2)-1);
pca_f5=pca(sub5(:,2:end),size(sub5,2)-1);
pca_f6=pca(sub6(:,2:end),size(sub6,2)-1);
outt0=blkdiag(pca_f1,pca_f2,pca_f3,pca_f4,pca_f5,pca_f6);
origi0=train90(2:end,2:end);
reordering=[f1(1,2:end) f2(1,2:end) f3(1,2:end) f4(1,2:end) f5(1,2:end) f6(1,2:end)];
storing=zeros(size(outt0,1),41);
for j=1:41;
    gi=reordering(j);
    storing(:,gi-1)=outt0(:,gi-1);
end
new_features0=origi0*(storing');
TestSet0=test10(:,2:end);
labelsss0=test10(:,1);
TrainingSet0=new_features0;
GroupTrain0=train90(2:end,1);
model0 = svmtrain(GroupTrain0, TrainingSet0, 's 0 t 1');
[predicted_label0] = svmpredict(labelsss0, TestSet0, model0);
stats0 = confusionmatStats(labelsss0,predicted_label0);
 end