Abstract:We study the matrix completion problem: an underlying matrix P is low rank, with incoherent singular vectors, and a random matrix A is equal to P on a (uniformly) random subset of entries of size dn . All other entries of A are equal to zero. The goal is to retrieve information on P from the observation of A . Let be the random matrix where each entry of A is multiplied by an independent -Bernoulli random variable with parameter 1/2. This paper is about when, how and why the non-Hermitian eigen-spectra of the matrices and captures more of the relevant information about the principal component structure of A than the eigen-spectra of and . We show that the eigenvalues of the asymmetric matrices and with modulus greater than a detection threshold are asymptotically equal to the eigenvalues of and and that the associated eigenvectors are aligned as well. The central surprise is that by intentionally inducing asymmetry and additional randomness via the matrix, we can extract more information than if we had worked with the singular value decomposition (SVD) of A . The associated detection threshold is asymptotically exact and is non-universal since it explicitly depends on the element-wise distribution of the underlying matrix P . We show that reliable, statistically optimal but not perfect matrix recovery, via a universal data-driven algorithm, is possible above this detection threshold using the information extracted from the asymmetric eigen-decompositions. Averaging the left and right eigenvectors provably improves estimation accuracy but not the detection threshold. Our results encompass the very sparse regime where d is of order 1 where matrix completion via the SVD of A fails or produces unreliable recovery. We define another variant of this asymmetric principal component analysis procedure that bypasses the randomization step and has a detection threshold that is smaller by a constant factor but with a computational cost that is larger by a polynomial factor of the number of observed entries. Both detection thresholds allow to go beyond the barrier due to the well-known information theoretical limit for exact matrix completion found in the literature.

A Fast Spectral Relaxation Approach to Matrix Completion Via Kronecker Products.

Accelerated singular value thresholding for matrix completion.

Matrix Completion and Low-Rank SVD via Fast Alternating Least Squares

Faster Matrix Completion Using Randomized SVD

Accelerated and Inexact Soft-Impute for Large-Scale Matrix and Tensor Completion

A Singular Value Thresholding Algorithm for Matrix Completion

A Fast Matrix Completion Method Based on Matrix Bifactorization and QR Decomposition

T-product factorization based method for matrix and tensor completion problems

Solving a low-rank factorization model for matrix completion by a nonlinear successive over-relaxation algorithm

Fast and Accurate Matrix Completion Via Truncated Nuclear Norm Regularization

Fast Methods for Denoising Matrix Completion Formulations, with Applications to Robust Seismic Data Interpolation

Singular Value Decomposition Approximation via Kronecker Summations for Imaging Applications

Solving Low-Rank Matrix Completion Problems Efficiently

Matrix Completion in the Unit Hypercube via Structured Matrix Factorization

Adaptive Rank-One Matrix Completion using Sum of Outer Products

Representing Images By Multiple Kronecker Product Sum

An Alternating Direction Algorithm for Matrix Completion with Nonnegative Factors

An Efficient Gauss--Newton Algorithm for Symmetric Low-Rank Product Matrix Approximations

Tensor Factorization for Low-Rank Tensor Completion.

Detection Thresholds in Very Sparse Matrix Completion

Robust Matrix Completion Based on Factorization and Truncated-Quadratic Loss Function