Mutual information-based method for selecting informative feature sets

作者：

Highlights：

• We evaluate the performance of eight different mutual information-based feature selection methods.

• Based on the analysis of the evaluation results, we propose a novel mutual information-based feature selection method.

• The effectiveness of the proposed method is demonstrated by the experimental results on UCI datasets and object recognition.

摘要

Highlights•We evaluate the performance of eight different mutual information-based feature selection methods.•Based on the analysis of the evaluation results, we propose a novel mutual information-based feature selection method.•The effectiveness of the proposed method is demonstrated by the experimental results on UCI datasets and object recognition.

论文关键词：Feature selection,Mutual information

论文评审过程：Received 23 April 2012, Revised 14 April 2013, Accepted 27 April 2013, Available online 22 May 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.04.021