Multi-label feature selection considering label supplementation

作者：

Highlights：

• We distinguish the different effects of label relationship on feature selection.

• The dynamic change of label relationships based on different features is introduced.

• Two multi-label feature selection methods based on information theory are proposed.

• Experiments on fifteen data sets show the effectiveness of the proposed methods.

摘要

•We distinguish the different effects of label relationship on feature selection.•The dynamic change of label relationships based on different features is introduced.•Two multi-label feature selection methods based on information theory are proposed.•Experiments on fifteen data sets show the effectiveness of the proposed methods.

论文关键词：Multi-label learning,Multi-label feature selection,Information theory,Label relationships

论文评审过程：Received 13 July 2020, Revised 31 January 2021, Accepted 28 June 2021, Available online 29 June 2021, Version of Record 10 July 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108137