Local discriminative based sparse subspace learning for feature selection
作者：
Highlights：
• The proposed model preserves the local discriminant structure and local geometric structure of the data simultaneously.
• It can not only improve the discriminative ability of the algorithm, but also utilize the local geometric structure information of the data.
• L1-norm is introduced to constrain the feature selection matrix.
• It can ensure the sparsity of the feature selection matrix and improve the algorithm's discrimination ability.
• The experimental results show that the proposed algorithm is more effective than the other five feature selection algorithms.
摘要
•The proposed model preserves the local discriminant structure and local geometric structure of the data simultaneously.•It can not only improve the discriminative ability of the algorithm, but also utilize the local geometric structure information of the data.•L1-norm is introduced to constrain the feature selection matrix.•It can ensure the sparsity of the feature selection matrix and improve the algorithm's discrimination ability.•The experimental results show that the proposed algorithm is more effective than the other five feature selection algorithms.
论文关键词：Local discriminant model,Subspace learning,Sparse constraint,Feature selection
论文评审过程：Received 23 January 2017, Revised 20 March 2019, Accepted 24 March 2019, Available online 25 March 2019, Version of Record 5 April 2019.
论文官网地址：https://doi.org/10.1016/j.patcog.2019.03.026
