Robust sparse and low-redundancy multi-label feature selection with dynamic local and global structure preservation

Yonghao Li, Liang Hu, Wanfu Gao

Abstract: Highlights•SLMDS uses the self-expression model to preserve global label correlations.•A high-quality dynamic graph is designed to preserve the local label correlations.•An inner product term is used to select low-redundant features.•An efficient optimization method with provable convergence is designed.