A fast Branch-and-Bound algorithm for U-curve feature selection
作者：
Highlights：
• We introduce a fast Branch-and-Bound algorithm for optimal feature selection.
• The proposed algorithm is based on a U-curve assumption on the cost function.
• Experiments show the algorithm to be robust to violation of the U-curve assumption.
• The algorithm is demonstrated by application to optimal imaging operator design.
• The algorithm is also demonstrated by application to classifier feature selection.
摘要
•We introduce a fast Branch-and-Bound algorithm for optimal feature selection.•The proposed algorithm is based on a U-curve assumption on the cost function.•Experiments show the algorithm to be robust to violation of the U-curve assumption.•The algorithm is demonstrated by application to optimal imaging operator design.•The algorithm is also demonstrated by application to classifier feature selection.
论文关键词：Branch-and-bound algorithm,Feature selection,U-curve assumption,W-operator design
论文评审过程：Received 21 June 2015, Revised 9 August 2017, Accepted 11 August 2017, Available online 18 August 2017, Version of Record 18 September 2017.
论文官网地址：https://doi.org/10.1016/j.patcog.2017.08.013
