SVM 自定义kernel的一个简单的code 例子

function [ result_pre, result_r, max_result ] = kernel_SVM( train_data,test_data,train_label, test_label)
%train_data  : m*m train-train kernel matrix
% test_data   n*m test-train kernel matrix
% train_label m*1 train label
% test_label  n*1 test label


r=1e-10;
for i=1:50
    train_data_exp =exp(-r*train_data);
    test_data_exp =exp(-r*test_data);
    train_data_exp = [(1:size(index_train,2))', train_data_exp];
    test_data_exp = [(1:size(index_test,2))', test_data_exp];
    model_precomputed = svmtrain(train_label, train_data_exp, '-s 1 -t 4 -n 0.02 -c 1.21 -g 1');
    [predict_label_P, accuracy_P, dec_values_P] = svmpredict(test_label, test_data_exp, model_precomputed);
    %          if (accuracy_P(1,1)>max(result_pre))
    result_pre(result_index)=accuracy_P(1,1);
    result_r(result_index)=r;
    result_index=result_index+1;
    %          end
    r=r*sqrt(10);
end
max_result = max(result_pre);


end