Embedded multi-label feature selection via orthogonal regression

Xueyuan Xu, Fulin Wei, Tianze Yu, Jinxin Lu, Aomei Liu, Li Zhuo, Feiping Nie, Xia Wu

Abstract: Highlights•The proposed method is based on an embedding approach.•The local label correlations (feature-label) are explored by orthogonal regression.•Global redundancy and relevancy optimization have been added in orthogonal regression.•A simple yet effective algorithm is utilized to obtain an optimal solution.