Multicategory angle-based large-margin classification

Biometrika. 2014 Sep;101(3):625-640. doi: 10.1093/biomet/asu017. Epub 2014 Jul 23.

Abstract

Large-margin classifiers are popular methods for classification. Among existing simultaneous multicategory large-margin classifiers, a common approach is to learn k different functions for a k-class problem with a sum-to-zero constraint. Such a formulation can be inefficient. We propose a new multicategory angle-based large-margin classification framework. The proposed angle-based classifiers consider a simplex-based prediction rule without the sum-to-zero constraint, and enjoy more efficient computation. Many binary large-margin classifiers can be naturally generalized for multicategory problems through the angle-based framework. Theoretical and numerical studies demonstrate the usefulness of the angle-based methods.

Keywords: Hard classification; Probability estimation; Soft classification; Support vector machine.