Can we use multisvm to classify more than 2 groups

My project is License Plate recognition. I have taken the images of license plate of Indian vehicles as database. Four different states and each state having 5 different cities is considered for database creation. So i am thinking to use SVM classifier to classify each state and city. Is it possible using MultiSVM classifier? IF YES, then How? If NO, then please suggest some alternative as a classifier?

Answer 1

[mizuki]

The answer is YES. There is a function "fitcecoc" in Statistics and Machine Learning Toolbox. As you know, originally SVM is a binary classifier, but fitcecoc function combines several SVMs.

There are mainly two ways to combine classifiers - called one-versus-one and one-versus-all. You can change the option if you want to.

About "how to," please see the documentation of fitcecoc. It is very simple. http://www.mathworks.com/help/releases/R2016a/stats/fitcecoc.html

Also, if you prefer GUI, use Classification Learner Apps, which you an find on Apps tab in MATLAB Toolstrip (Menubar).