Robust L1-norm two-dimensional collaborative representation-based projection for dimensionality reduction
作者：
Highlights：
• Introducing L1-norm into CRP and a novel model for dimensionality reduction, named L1-2DCRP, is proposed to make it more robustness to outliers.
• The samples are extended form vectors to matrices.
• A simple but effective iterative algorithm with convergence proved is proposed to solve L1-2DCRP.
• Unlike the gradient used in many existing approaches for L1 norm optimization problem, a modified gradient is utilized and with a promising performance.
摘要
•Introducing L1-norm into CRP and a novel model for dimensionality reduction, named L1-2DCRP, is proposed to make it more robustness to outliers.•The samples are extended form vectors to matrices.•A simple but effective iterative algorithm with convergence proved is proposed to solve L1-2DCRP.•Unlike the gradient used in many existing approaches for L1 norm optimization problem, a modified gradient is utilized and with a promising performance.
论文关键词：Collaborative representation-based projection (CRP),L1-2DCRP,L1-norm,Face recognition,Dimensionality reduction
论文评审过程：Received 10 June 2019, Revised 29 October 2019, Accepted 1 November 2019, Available online 8 November 2019, Version of Record 13 November 2019.
论文官网地址：https://doi.org/10.1016/j.image.2019.115684
