Question

I wish to implement multi-class SVM using one-vs-one method on fisheriris data in MATLAB.

How can I do so ?

Answer 1

You can use the libsvm library for SVMs.

You can train an svm with:

model = svmtrain(features, labels);

and make predictions:

predictions = svmpredict(features);

By default, if there are multiple values in ‘labels’, libsvm does 1vs1 classification.