Embedded feature selection accounting for unknown data heterogeneity
作者：
Highlights：
• Data heterogeneity leads to spurious classification and feature selection results.
• Our embedded feature selection method can account for unknown data heterogeneity.
• Sparse optimal scoring on the adjusted data is proposed for multi-class classification.
• Effective proximal gradient update rules are developed to find optimal solutions.
• Our method outperforms the state-of-the-arts on synthetic data and three benchmark image datasets.
摘要
•Data heterogeneity leads to spurious classification and feature selection results.•Our embedded feature selection method can account for unknown data heterogeneity.•Sparse optimal scoring on the adjusted data is proposed for multi-class classification.•Effective proximal gradient update rules are developed to find optimal solutions.•Our method outperforms the state-of-the-arts on synthetic data and three benchmark image datasets.
论文关键词：Feature selection,Data heterogeneity,Embedded method,Sparse optimal scoring
论文评审过程：Received 11 July 2018, Revised 26 September 2018, Accepted 5 November 2018, Available online 6 November 2018, Version of Record 10 November 2018.
论文官网地址：https://doi.org/10.1016/j.eswa.2018.11.006
