SCLS: Multi-label feature selection based on scalable criterion for large label set

作者：

Highlights：

• A multi-label feature selection method for multi-label classification is proposed.

• We propose a new scalable relevance evaluation process for feature evaluation.

• The proposed method is designed to use a simpler dependency calculation process.

• An effective approximation for the relevance evaluation is devised.

摘要

Highlights•A multi-label feature selection method for multi-label classification is proposed.•We propose a new scalable relevance evaluation process for feature evaluation.•The proposed method is designed to use a simpler dependency calculation process.•An effective approximation for the relevance evaluation is devised.

论文关键词：Machine learning,Multi-label learning,Multi-label feature selection,Relevance evaluation,Conditional relevance

论文评审过程：Received 12 August 2016, Revised 13 December 2016, Accepted 9 January 2017, Available online 10 January 2017, Version of Record 12 March 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.01.014