Accelerated Alternating Projections for Robust Principal Component Analysis

Abstract

We study robust PCA for the fully observed setting, which is about separating a low rank matrix $\mathbf{L}$ and a sparse matrix $\mathbf{S}$ from their sum $\mathbf{D}=\mathbf{L}+\mathbf{S}$. In this paper, a new algorithm, dubbed accelerated alternating projections, is introduced for robust PCA which significantly improves the computational efficiency of the existing alternating projections proposed in [Netrapalli, Praneeth, et al., 2014] when updating the low rank factor. The acceleration is achieved by first projecting a matrix onto some low dimensional subspace before obtaining a new estimate of the low rank matrix via truncated SVD. Exact recovery guarantee has been established which shows linear convergence of the proposed algorithm. Empirical performance evaluations establish the advantage of our algorithm over other state-of-the-art algorithms for robust PCA.

Cited by in corpus (10)

• Rapid Robust Principal Component Analysis: CUR Accelerated Inexact Low Rank Estimation
• Accelerated Structured Alternating Projections for Robust Spectrally Sparse Signal Recovery
• Learned Robust PCA: A Scalable Deep Unfolding Approach for High-Dimensional Outlier Detection
• Mode-wise Tensor Decompositions: Multi-dimensional Generalizations of CUR Decompositions
• Alternating projections gridless covariance-based estimation for DOA
• Bridging Convex and Nonconvex Optimization in Robust PCA: Noise, Outliers, and Missing Data
• Sparse Plus Low Rank Matrix Decomposition: A Discrete Optimization Approach
• Fast algorithms for robust principal component analysis with an upper bound on the rank
• Fast Robust Tensor Principal Component Analysis via Fiber CUR Decomposition
• Fast Alternating Projections on Manifolds Based on Tangent Spaces