Please use this identifier to cite or link to this item:

Coherence functions for multicategory margin-based classification methods

Authors Zhang, Zhihua HKUST affiliated (currently or previously)
Jordan, Michael
Li, Wu-Jun HKUST affiliated (currently or previously)
Yeung, Dit Yan View this author's profile
Issue Date 2009
Source Twelfth International Conference on Artificial Intelligence and Statistics (AISTATS), Clearwater Beach, Florida, USA , 16-18 April 2009, p. 647-654
Summary Margin-based classification methods are typically devised based on a majorization-minimization procedure, which approximately solves an otherwise intractable minimization problem defined with the 0-l loss. The extension of such methods from the binary classification setting to the more general multicategory setting turns out to be non-trivial. In this paper, our focus is to devise margin-based classification methods that can be seamlessly applied to both settings, with the binary setting simply as a special case. In particular, we propose a new majorization loss function that we call the coherence function, and then devise a new multicategory margin-based boosting algorithm based on the coherence function. Analogous to deterministic annealing, the coherence function is characterized by a temperature factor. It is closely related to the multinomial log-likelihood function and its limit at zero temperature corresponds to a multicategory hinge loss function.
Rights Reprinted with permission from MIT Press
Language English
Format Conference paper
Files in this item:
File Description Size Format
coher.pdf 530063 B Adobe PDF