Unsupervised feature selection using an improved version of Differential Evolution

作者：

Highlights：

• Unsupervised feature selection approach using improved version of Differential Evolution technique.

• Normalized mutual information score is utilized for computing similarity and dissimilarity.

• Superiority over three existing methods is established for eight data sets.

摘要

•Unsupervised feature selection approach using improved version of Differential Evolution technique.•Normalized mutual information score is utilized for computing similarity and dissimilarity.•Superiority over three existing methods is established for eight data sets.

论文关键词：Pattern recognition,Unsupervised feature selection,Mutual information,Normalized mutual information,Differential Evolution

论文评审过程：Available online 16 December 2014.

论文官网地址：https://doi.org/10.1016/j.eswa.2014.12.010