Abstract
The construction of multiclass classifiers from binary classifiers is studied in this paper, and performance is quantified by the regret, defined with respect to the Bayes optimal log-loss. We start by proving that the regret of the well known One vs. All (OVA) method is upper bounded by the sum of the regrets of its constituent binary classifiers. We then present a new method called Conditional OVA (COVA), and prove that its regret is given by the weighted sum of the regrets corresponding to the constituent binary classifiers. Lastly, we present a method termed Leveraged COVA (LCOVA), designated to reduce the regret of a multiclass classifier by breaking it down to independently optimized binary classifiers.
Original language | English |
---|---|
Title of host publication | 2021 IEEE International Symposium on Information Theory, ISIT 2021 - Proceedings |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
Pages | 2435-2440 |
Number of pages | 6 |
ISBN (Electronic) | 9781538682098 |
DOIs | |
State | Published - 12 Jul 2021 |
Event | 2021 IEEE International Symposium on Information Theory, ISIT 2021 - Virtual, Melbourne, Australia Duration: 12 Jul 2021 → 20 Jul 2021 |
Publication series
Name | IEEE International Symposium on Information Theory - Proceedings |
---|---|
Volume | 2021-July |
ISSN (Print) | 2157-8095 |
Conference
Conference | 2021 IEEE International Symposium on Information Theory, ISIT 2021 |
---|---|
Country/Territory | Australia |
City | Virtual, Melbourne |
Period | 12/07/21 → 20/07/21 |
Bibliographical note
Publisher Copyright:© 2021 IEEE.