Active k-labelsets ensemble for multi-label classification

作者：

Highlights：

• ACkEL is proposed for multi-label classification by improving RAkEL.

• Separability and balance level are used to evaluate the quality of a label subset.

• ACkEL can be realized in both disjoint mode and overlapping mode.

• The outcome indicates that ACkEL is able to produce promising results.

摘要

•ACkEL is proposed for multi-label classification by improving RAkEL.•Separability and balance level are used to evaluate the quality of a label subset.•ACkEL can be realized in both disjoint mode and overlapping mode.•The outcome indicates that ACkEL is able to produce promising results.

论文关键词：Multi-label learning,k-Labelsets Ensemble,Label powerset,Separability

论文评审过程：Received 17 December 2019, Revised 14 July 2020, Accepted 7 August 2020, Available online 8 August 2020, Version of Record 21 August 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107583