Sparse robust subspace learning via boolean weight

Sisi Wang, Feiping Nie, Zheng Wang, Rong Wang, Xuelong Li

Abstract: Highlights•A robust feature selection method based on reconstruction and variance is proposed.•Designing boolean weight to fundamentally eliminate outliers.•Designing an algorithm to solve ℓ2,0-norm constrained problem of feature selection.•Our algorithm has significant advantages in practical applications.