Evaluation of feature selection methods based on artificial neural network weights

作者：

Highlights：

• A new methodology to evaluate the stability of WBFS is proposed.

• A voting approach was used to compare different importance rankings.

• Experiments on simulated and real-world datasets.

• Existing WBFS methods are not stable on real-world datasets.

摘要

•A new methodology to evaluate the stability of WBFS is proposed.•A voting approach was used to compare different importance rankings.•Experiments on simulated and real-world datasets.•Existing WBFS methods are not stable on real-world datasets.

论文关键词：Relative importance,Feature selection,Garson,Olden,Importance ranking,Neural networks

论文评审过程：Received 28 March 2020, Revised 4 August 2020, Accepted 10 November 2020, Available online 17 November 2020, Version of Record 8 December 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114312