Understanding the apparent superiority of over-sampling through an analysis of local information for class-imbalanced data

作者：

Highlights：

• Distribution samples is computed in both classes on several imbalanced datasets.

• Resampling techniques are explored by analyzing the distribution of samples.

• Over-sampling techniques provide a higher/lower proportion of safe/unsafe samples.

摘要

•Distribution samples is computed in both classes on several imbalanced datasets.•Resampling techniques are explored by analyzing the distribution of samples.•Over-sampling techniques provide a higher/lower proportion of safe/unsafe samples.

论文关键词：Class imbalance,Sample types,Resampling,Local neighborhood

论文评审过程：Received 4 February 2019, Revised 12 October 2019, Accepted 13 October 2019, Available online 14 October 2019, Version of Record 8 September 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.113026