A Unified Framework for Outlier-Robust PCA-like Algorithms

We propose a unified framework for making a wide range of PCA-like algorithms – including the standard PCA, sparse PCA and non-negative sparse PCA, etc. – robust when facing a constant fraction of arb