Feature selection in possibilistic modeling
作者：
Highlights：
• We present a feature selection strategy for possibilitic modeling.
• We examine feature importance within one class and for classes׳s discrimination.
• resolving conflict between feature׳s importance using possibility uncertainty degree.
• Proposed strategy is able to handle data imperfection and gives good performances.
• Proposed strategy gives better and reliable results on any data set.
摘要
Highlights•We present a feature selection strategy for possibilitic modeling.•We examine feature importance within one class and for classes׳s discrimination.•resolving conflict between feature׳s importance using possibility uncertainty degree.•Proposed strategy is able to handle data imperfection and gives good performances.•Proposed strategy gives better and reliable results on any data set.
论文关键词：Feature selection,Shapley index,Possibility theory,Possibility distribution uncertainty,Class representation,Classes׳s discrimination
论文评审过程：Received 28 July 2014, Revised 24 January 2015, Accepted 14 March 2015, Available online 9 April 2015, Version of Record 16 July 2015.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.03.015
