A scalable saliency-based feature selection method with instance-level information

作者：

Highlights：

• Most feature selection techniques are unable to provide instance-level information.

• A novel method based on deep-learning saliency techniques is proposed.

• It can be used in any architecture trained using gradient descent technique.

• We successfully tested our method on different datasets, getting encouraging results.

摘要

•Most feature selection techniques are unable to provide instance-level information.•A novel method based on deep-learning saliency techniques is proposed.•It can be used in any architecture trained using gradient descent technique.•We successfully tested our method on different datasets, getting encouraging results.

论文关键词：Feature selection,Deep learning,Saliency

论文评审过程：Received 13 May 2019, Revised 28 November 2019, Accepted 30 November 2019, Available online 6 December 2019, Version of Record 24 February 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2019.105326