Robust principal component analysis via optimal mean by joint ℓ 2,1 and Schatten p -norms minimization

摘要: Abstract Since principal component analysis (PCA) is sensitive to corrupted variables or observations that affect its performance and applicability in real scenarios, some convex robust PCA methods have been developed enhance the robustness of PCA. However, most them neglect optimal mean calculation problem. They center data with calculated by l2-norm, which incorrect because l1-norm objective function used following steps. In this paper, we consider a novel method can pursue remove outliers, exactly recover low-rank matrix calculate mean. Specifically, propose an optimization model constituted l2,1-norm based loss Schatten p-norm regularization term. The aims suppress singular values reconstructed at smaller p (0