A novel framework for parsimonious multivariate analysis
作者：
Highlights：
• Parsimonious: all extracted features come from a compact subset of the input ones.
• Feature selection: spot consistencies in members of ensemble done with bootstrapping.
• Asses individual relevance of variables to lead final parsimonious feature extraction.
• Improvement in classification accuracy and interpretation of the results.
摘要
•Parsimonious: all extracted features come from a compact subset of the input ones.•Feature selection: spot consistencies in members of ensemble done with bootstrapping.•Asses individual relevance of variables to lead final parsimonious feature extraction.•Improvement in classification accuracy and interpretation of the results.
论文关键词：Feature selection,Dimensionality reduction,Multivariate analysis,Principal component analysis,Canonical correlation analysis,Orthonormalized Partial Least Squares
论文评审过程：Received 20 September 2016, Revised 25 March 2017, Accepted 1 June 2017, Available online 3 June 2017, Version of Record 13 June 2017.
论文官网地址：https://doi.org/10.1016/j.patcog.2017.06.004
