Capped norm based discriminant robust regression learning

Ning Liu, Zhihui Lai, Junhong Zhang, Can Gao, Heng Kong

Abstract: Highlights•CNDRR is proposed for robust feature extraction.•In CNDRR, latent information is utilized, and the discrimination ability is further improved.•The underlying relevant structure between classes is uncovered in CNDRR.•An alternating iterative algorithm is proposed to solve the CNDRR optimization problem.•Experiments are conducted on eight well-known databases to show the proposed CNDRR’s superiority against other methods.