Orthogonal least squares based fast feature selection for linear classification

Highlights：

• An orthogonal least squares based feature selection method is then proposed and it is shown that the method has speed advantages when applied for the greedy search.

• A comparison of the proposed method with the mutual information based feature selection methods and the embedded methods shows that the proposed method is always in the top 5 among the 12 candidate methods.

摘要

•Based on orthogonal least squares, the novel squared orthogonal correlation coefficients are defined and their relationship with canonical correlation coefficient and linear discriminant analysis is revealed.•An orthogonal least squares based feature selection method is then proposed and it is shown that the method has speed advantages when applied for the greedy search.•A comparison of the proposed method with the mutual information based feature selection methods and the embedded methods shows that the proposed method is always in the top 5 among the 12 candidate methods.

论文关键词：Feature selection,Orthogonal least squares,Canonical correlation analysis,Linear discriminant analysis,Multi-label,Multivariate time series,Feature interaction

论文评审过程：Received 17 November 2020, Revised 30 September 2021, Accepted 1 November 2021, Available online 2 November 2021, Version of Record 13 November 2021.