The Augmented Lagrange Multiplier Method for Exact Recovery of Corrupted Low-Rank Matrices

TL;DR

This paper introduces an augmented Lagrange multiplier method for efficiently solving the Robust PCA problem, enabling exact recovery of low-rank matrices with corrupted entries, with faster convergence and higher accuracy than previous algorithms.

Contribution

It extends ALM analysis to non-smooth objectives and demonstrates its effectiveness for Robust PCA and matrix completion, improving speed and precision.

Findings

Algorithms are over five times faster than previous methods.

Achieves higher precision with less memory usage.

Proves global convergence conditions for inexact ALM.

Abstract

This paper proposes scalable and fast algorithms for solving the Robust PCA problem, namely recovering a low-rank matrix with an unknown fraction of its entries being arbitrarily corrupted. This problem arises in many applications, such as image processing, web data ranking, and bioinformatic data analysis. It was recently shown that under surprisingly broad conditions, the Robust PCA problem can be exactly solved via convex optimization that minimizes a combination of the nuclear norm and the $\ell^1$-norm . In this paper, we apply the method of augmented Lagrange multipliers (ALM) to solve this convex program. As the objective function is non-smooth, we show how to extend the classical analysis of ALM to such new objective functions and prove the optimality of the proposed algorithms and characterize their convergence rate. Empirically, the proposed new algorithms can be more than five times faster than the previous state-of-the-art algorithms for Robust PCA, such as the accelerated proximal gradient (APG) algorithm. Moreover, the new algorithms achieve higher precision, yet being less storage/memory demanding. We also show that the ALM technique can be used to solve the (related but somewhat simpler) matrix completion problem and obtain rather promising results too. We further prove the necessary and sufficient condition for the inexact ALM to converge globally. Matlab code of all algorithms discussed are available at http://perception.csl.illinois.edu/matrix-rank/home.html