Abstract:Many applications require recovering a matrix of minimal rank within an affine constraint set, with matrix completion a notable special case. Because the problem is NP-hard in general, it is common to replace the matrix rank with the nuclear norm, which acts as a convenient convex surrogate. While elegant theoretical conditions elucidate when this replacement is likely to be successful, they are highly restrictive and convex algorithms fail when the ambient rank is too high or when the constraint set is poorly structured. Nonconvex alternatives fare somewhat better when carefully tuned; however, convergence to locally optimal solutions remains a continuing source of failure. Against this backdrop, we derive a deceptively simple and parameter-free probabilistic PCA-like algorithm that is capable, over a wide battery of empirical tests, of successful recovery even at the theoretical limit where the number of measurements equals the degrees of freedom in the unknown low-rank matrix. Somewhat surprisingly, this is possible even when the affine constraint set is highly ill-conditioned. While proving general recovery guarantees remains evasive for nonconvex algorithms, Bayesian-inspired or otherwise, we nonetheless show conditions whereby the underlying cost function has a unique stationary point located at the global optimum; no existing cost function we are aware of satisfies this property. The algorithm has also been successfully deployed on a computer vision application involving image rectification and a standard collaborative filtering benchmark.

A Counterexample for the Validity of Using Nuclear Norm as a Convex Surrogate of Rank

Recovery of Low-Rank Matrices Based on the Rank Null Space Properties.

Nonconvex Nonsmooth Low-Rank Minimization Via Iteratively Reweighted Nuclear Norm

The Bounds of Restricted Isometry Constants for Low Rank Matrices Recovery

Guaranteed Minimum-Rank Solutions of Linear Matrix Equations via Nuclear Norm Minimization

On the Schatten P-Quasi-norm Minimization for Low-Rank Matrix Recovery

Low Rank Matrix Recovery with Adversarial Sparse Noise

A robust low-rank matrix completion based on truncated nuclear norm and Lp-norm

Exploring Algorithmic Limits of Matrix Rank Minimization under Affine Constraints

Rank Aggregation via Nuclear Norm Minimization

Nuclear Norm Minus Frobenius Norm Minimization with Rank Residual Constraint for Image Denoising

Matrix completion by Truncated Nuclear Norm Regularization

Lower bounds on nonnegative rank via nonnegative nuclear norms

Efficient Low-rank Identification via Accelerated Iteratively Reweighted Nuclear Norm Minimization

Robust Recovery of Low-Rank Matrices Via Non-Convex Optimization

Finding approximately rank-one submatrices with the nuclear norm and l1 norm

Robust PCA via Nonconvex Rank Approximation

Low solution rank of the matrix LASSO under RIP with consequences for rank-constrained algorithms

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

Restricted -Isometry Properties of Nonconvex Matrix Recovery

A Nonconvex Relaxation Approach for Rank Minimization Problems