A Bias Trick for Centered Robust Principal Component Analysis
arXiv:1911.08024v11 citations
AI Analysis
This addresses a specific technical issue in RPCA for researchers and practitioners in machine learning, but it is incremental as it focuses on improving centering within an existing framework.
The paper tackles the problem of centering non-outliers in Robust Principal Component Analysis (RPCA) by introducing a 'bias trick' that automates this process, resulting in the first RPCA algorithm that is optimal with respect to centering.
Outlier based Robust Principal Component Analysis (RPCA) requires centering of the non-outliers. We show a "bias trick" that automatically centers these non-outliers. Using this bias trick we obtain the first RPCA algorithm that is optimal with respect to centering.