Abstract:This paper introduces a novel algorithm to approximate the matrix with minimum nuclear norm among all matrices obeying a set of convex constraints. This problem may be understood as the convex relaxation of a rank minimization problem and arises in many important applications as in the task of recovering a large matrix from a small subset of its entries (the famous Netflix problem). Off-the-shelf algorithms such as interior point methods are not directly amenable to large problems of this kind with over a million unknown entries. This paper develops a simple first-order and easy-to-implement algorithm that is extremely efficient at addressing problems in which the optimal solution has low rank. The algorithm is iterative, produces a sequence of matrices $\{\boldsymbol{X}^k,\boldsymbol{Y}^k\}$, and at each step mainly performs a soft-thresholding operation on the singular values of the matrix $\boldsymbol{Y}^k$. There are two remarkable features making this attractive for low-rank matrix completion problems. The first is that the soft-thresholding operation is applied to a sparse matrix; the second is that the rank of the iterates $\{\boldsymbol{X}^k\}$ is empirically nondecreasing. Both these facts allow the algorithm to make use of very minimal storage space and keep the computational cost of each iteration low. On the theoretical side, we provide a convergence analysis showing that the sequence of iterates converges. On the practical side, we provide numerical examples in which $1,000\times1,000$ matrices are recovered in less than a minute on a modest desktop computer. We also demonstrate that our approach is amenable to very large scale problems by recovering matrices of rank about 10 with nearly a billion unknowns from just about 0.4% of their sampled entries. Our methods are connected with the recent literature on linearized Bregman iterations for $\ell_1$ minimization, and we develop a framework in which one can understand these algorithms in terms of well-known Lagrange multiplier algorithms.

Greedy Matrix Completion with Fitting Error and Rank Iterative Minimization

A Nearly Unbiased Matrix Completion Approach.

Fast and Accurate Matrix Completion Via Truncated Nuclear Norm Regularization

Adaptive Rank-One Matrix Completion using Sum of Outer Products

Locally Linear Approximation Approach for Incomplete Data.

Matrix completion based on non-convex low rank approximation

Solving Low-Rank Matrix Completion Problems Efficiently

Matrix completion algorithms with optimal phase transition

A Novel Nonconvex Relaxation Approach to Low-Rank Matrix Completion of Inexact Observed Data

A Nonconvex Method to Low-Rank Matrix Completion

A Riemannian rank-adaptive method for low-rank matrix completion

Exploring Algorithmic Limits of Matrix Rank Minimization under Affine Constraints

Matrix completion by Truncated Nuclear Norm Regularization

A Singular Value Thresholding Algorithm for Matrix Completion

Low rank matrix completion by alternating steepest descent methods

Partial Matrix Completion

Robust Matrix Completion Via Alternating Projection.

A New Retraction For Accelerating The Riemannian Three-Factor Low-Rank Matrix Completion Algorithm

Low Rank Matrix Completion via Robust Alternating Minimization in Nearly Linear Time

Truncated quadratic norm minimization for bilinear factorization based matrix completion

Exact Error in Matrix Completion: Approximately Low-Rank Structures and Missing Blocks