Comprehensive comparative study of multi-label classification methods

作者：

Highlights：

• A comprehensive empirical investigation of multi-label classification methods.

• A strict and robust evaluation methodology for large-scale benchmarking.

• Evaluated 26 methods across 42 benchmark datasets using 20 evaluation measures.

• Best overall methods are RFPCT, RFDTBR, ECCJ48, EBRJ48, and AdaBoost. MH.

摘要

•A comprehensive empirical investigation of multi-label classification methods.•A strict and robust evaluation methodology for large-scale benchmarking.•Evaluated 26 methods across 42 benchmark datasets using 20 evaluation measures.•Best overall methods are RFPCT, RFDTBR, ECCJ48, EBRJ48, and AdaBoost. MH.

论文关键词：Multi-label classification,Benchmarking machine learning methods,Performance estimation,Evaluation measures

论文评审过程：Received 2 February 2022, Revised 8 April 2022, Accepted 8 April 2022, Available online 21 April 2022, Version of Record 10 May 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117215