error using ovr for multi class svm

5 次查看（过去 30 天）

Fateme Jalali 2016-8-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/299650-error-using-ovr-for-multi-class-svm

编辑： Walter Roberson 2016-8-15

Hello,I am using libsvm to classify data into 60 classes.size of my train matrix is 360*200.360 is the number of observstion and 200 is the number of features. I use libsvm from here: https://www.researchgate.net/post/How_can_I_use_Libsvm_in_matlab_for_multi-class_SVM. I recieve this error:

Index exceeds matrix dimensions.
Error in OVR (line 48)
    trainData = data(trainIndex,:);
Error in sample (line 99)
[predictedLabel, decisValueWinner, totalAccuracy, confusionMatrix, order] = OVR(label,
ndata, runCVIndex, Ncv_classif, optionClassif);