Multiple Classifier Systems

Volume 3541 of the series Lecture Notes in Computer Science pp 278-285

Which Is the Best Multiclass SVM Method? An Empirical Study

  • Kai-Bo DuanAffiliated withBioInformatics Research Centre, Nanyang Technological University
  • , S. Sathiya KeerthiAffiliated withYahoo! Research Labs

* Final gross prices may vary according to local VAT.

Get Access


Multiclass SVMs are usually implemented by combining several two-class SVMs. The one-versus-all method using winner-takes-all strategy and the one-versus-one method implemented by max-wins voting are popularly used for this purpose. In this paper we give empirical evidence to show that these methods are inferior to another one-versus-one method: one that uses Platt’s posterior probabilities together with the pairwise coupling idea of Hastie and Tibshirani. The evidence is particularly strong when the training dataset is sparse.