Feature selection with redundancy-complementariness dispersion
作者：
Highlights：
• We use inter-correlation of features to represent redundancy and complementariness.
• We add a modification item for feature complementariness in the evaluation function.
• Redundancy-complementariness dispersion is used to address the interference effect.
摘要
•We use inter-correlation of features to represent redundancy and complementariness.•We add a modification item for feature complementariness in the evaluation function.•Redundancy-complementariness dispersion is used to address the interference effect.
论文关键词：Classification,Feature selection,Relevance,Redundancy,Pairwise approximation,Redundancy-complementariness dispersion
论文评审过程：Received 2 February 2015, Revised 18 May 2015, Accepted 12 July 2015, Available online 18 July 2015, Version of Record 19 October 2015.
论文官网地址：https://doi.org/10.1016/j.knosys.2015.07.004
