Ensemble methods for multi-label classification

作者：

Highlights：

• New framework of multi-label ensemble classification algorithms is introduced.

• The set covering problem (SCP) is used to construct the labelsets.

• Constraints are incorporated in the construction of the labelsets.

• Theoretical and empirical analysis and comparison are provided.

• The proposed methods outperform RAKEL and other state-of-the-art algorithms.

摘要

•New framework of multi-label ensemble classification algorithms is introduced.•The set covering problem (SCP) is used to construct the labelsets.•Constraints are incorporated in the construction of the labelsets.•Theoretical and empirical analysis and comparison are provided.•The proposed methods outperform RAKEL and other state-of-the-art algorithms.

论文关键词：Multi-label classification,Ensemble learning

论文评审过程：Available online 16 June 2014.

论文官网地址：https://doi.org/10.1016/j.eswa.2014.06.015