The Convexity and Design of Composite Multiclass Losses
Machine Learning
2012-06-22 v1 Machine Learning
Abstract
We consider composite loss functions for multiclass prediction comprising a proper (i.e., Fisher-consistent) loss over probability distributions and an inverse link function. We establish conditions for their (strong) convexity and explore the implications. We also show how the separation of concerns afforded by using this composite representation allows for the design of families of losses with the same Bayes risk.
Keywords
Cite
@article{arxiv.1206.4663,
title = {The Convexity and Design of Composite Multiclass Losses},
author = {Mark Reid and Robert Williamson and Peng Sun},
journal= {arXiv preprint arXiv:1206.4663},
year = {2012}
}
Comments
ICML2012