Composite multiclass losses
Date
2016
Authors
Williamson, Robert
Vernet, Elodie
Reid, Mark
Journal Title
Journal ISSN
Volume Title
Publisher
MIT Press
Abstract
We consider loss functions for multiclass prediction problems. We show when a multiclass loss can be expressed as a “proper composite loss”, which is the composition of a proper loss and a link function. We extend existing results for binary losses to multiclass losses. We subsume results on “classification calibration” by relating it to properness. We determine the stationarity condition, Bregman representation, order-sensitivity, and quasi-convexity of multiclass proper losses. We then characterise the existence and uniqueness of the composite representation formulti class losses. We show how the composite representation is related to other core properties of a loss: mixability, admissibility and (strong) convexity of multiclass losses which we characterise in terms of the Hessian of the Bayes risk. We show that the simple integral representation for binary proper losses can not be extended to multiclass losses but offer concrete guidance regarding how to design different loss functions. The conclusion drawn from these results is that the proper composite representation is a natural and convenient tool for the design of multiclass loss functions.
Description
Keywords
Citation
Collections
Source
Journal of Machine Learning Research
Type
Journal article
Book Title
Entity type
Access Statement
Open Access