Maximum-relevance and maximum-diversity of positive ranks: A novel feature selection method

作者：

Highlights：

• Ranks of positive samples are used to define relevance of features.

• Diversity of the positive ranks is used to define diversity of features.

• Combination of relevance and diversity is defined as a feature selection metric.

• Better performance scores than several mutual information-based methods are reported.

摘要

•Ranks of positive samples are used to define relevance of features.•Diversity of the positive ranks is used to define diversity of features.•Combination of relevance and diversity is defined as a feature selection metric.•Better performance scores than several mutual information-based methods are reported.

论文关键词：Feature selection,Ranks of positives,Relevance,Redundancy,Diversity,Multivariate filter

论文评审过程：Received 4 October 2018, Revised 28 April 2020, Accepted 29 April 2020, Available online 11 May 2020, Version of Record 20 May 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113499