Generation of compound features based on feature interaction for classification
作者：
Highlights：
• Novel compound feature generation method using mutual information is proposed.
• It creates new semi-features by transforming significantly interacting features.
• Removal of irrelevant features is done only after learning semi-features.
• The method able to produce maximum informative and less redundant compound features.
• Accuracy of the method is compared with state-of-the-art DR methods using multiple classifiers.
摘要
•Novel compound feature generation method using mutual information is proposed.•It creates new semi-features by transforming significantly interacting features.•Removal of irrelevant features is done only after learning semi-features.•The method able to produce maximum informative and less redundant compound features.•Accuracy of the method is compared with state-of-the-art DR methods using multiple classifiers.
论文关键词：Feature extraction,Feature selection,Compound features,Semi-features,Information theory,Feature interaction,Mutual information
论文评审过程：Received 19 November 2017, Revised 26 April 2018, Accepted 26 April 2018, Available online 30 April 2018, Version of Record 5 May 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.04.033
