Ranking-based instance selection for pattern classification

作者：

Highlights：

• Three novel algorithms of the RIS family are proposed.

• RIS employs a ranking that selects the best instances in terms of classification.

• The ranking implies that borderline and noisy instances have low priority.

• RIS obtains promising accuracy and reduction rates when compared with the literature.

摘要

•Three novel algorithms of the RIS family are proposed.•RIS employs a ranking that selects the best instances in terms of classification.•The ranking implies that borderline and noisy instances have low priority.•RIS obtains promising accuracy and reduction rates when compared with the literature.

论文关键词：Instance selection,Ranking,Instance-based learning,k-nearest neighbor,Classification

论文评审过程：Received 16 May 2019, Revised 6 January 2020, Accepted 31 January 2020, Available online 7 February 2020, Version of Record 19 February 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113269