Study of data transformation techniques for adapting single-label prototype selection algorithms to multi-label learning

作者：

Highlights：

• Data transformation methods for instance selection (IS) in multi-label problems is investigated.

• The approach allows the adaptation to multi-label of IS methods for single-label.

• The adaptation of LSSm algorithm using binary relevance shows competitive results.

• At the moment, the IS is not as advantageous for multi-label as it is for single-label.

摘要

•Data transformation methods for instance selection (IS) in multi-label problems is investigated.•The approach allows the adaptation to multi-label of IS methods for single-label.•The adaptation of LSSm algorithm using binary relevance shows competitive results.•At the moment, the IS is not as advantageous for multi-label as it is for single-label.

论文关键词：Multi-label classification,Prototype selection,Binary relevance,Label powerset,RAkEL

论文评审过程：Received 5 September 2017, Revised 26 April 2018, Accepted 15 May 2018, Available online 26 May 2018, Version of Record 26 May 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.05.017