Can we use SVM for multi class classification?

Hello Syed,

Good to hear from you!

Yes, SVM can be used for multi-class classification. While SVM is fundamentally a binary classifier, it can be extended for multi-class classification using strategies such as:

1) One-vs-One: This involves training a separate classifier for every pair of classes. If you have N classes, you'll need to train N×(N−1)/2 classifiers. During prediction, the class that wins the most duels is chosen.
2) One-vs-Rest (OvR) or One-vs-All (OvA): This strategy trains a single classifier per class, with the samples of that class as positive samples and all other samples as negatives. During prediction, the classifier with the highest decision function score is picked as the class.
3) Error-Correcting Output Codes (ECOC): This more complex method represents the classes in a binary code (with each bit being the output of a binary classifier) and trains classifiers to separate these codes. During prediction, the class whose code is closest to the output code is chosen.

Hope this helps!

Best,

Ned