Designing multi-label classifiers that maximize F measures: State of the art
详细信息 查看全文
- 作者：Ignazio Pillai pillai@diee.unica.it" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Giorgio Fumera ; fumera@diee.unica.it" class="auth_mail" title="E-mail the corresponding author ; Author Vitae ; Fabio Roli roli@diee.unica.it" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae
- 关键词：Multi-label classification ; F measure ; Learning algorithms ; Empirical utility maximization ; Decision-theoretic approach
- 刊名：Pattern Recognition
- 出版年：2017
- 出版时间：January 2017
- 年：2017
- 卷：61
- 期：Complete
- 页码：394-404
- 全文大小：688 K
文摘
- label">•
We survey classification algorithms that maximize F measures.
- label">•
We consider the empirical utility maximization and decision-theoretic approaches.
- label">•
We consider first the single-label F measure.
- label">•
We then consider the multi-label instance-wise, macro- and micro-averaged F.