Completing Any Low-rank Matrix, Provably

TL;DR

This paper proves that any low-rank matrix can be exactly recovered from a small number of samples if the sampling is biased according to leverage scores, extending matrix completion beyond incoherent matrices.

Contribution

It introduces a novel biased sampling scheme based on leverage scores that guarantees exact recovery of any low-rank matrix, and provides methods for practical implementation without prior leverage score knowledge.

Findings

Exact recovery from O(nr log^2 n) samples under biased sampling.

Biased sampling based on leverage scores is nearly necessary for recovery.

Weighted nuclear norm minimization outperforms unweighted methods under non-uniform sampling.

Abstract

Matrix completion, i.e., the exact and provable recovery of a low-rank matrix from a small subset of its elements, is currently only known to be possible if the matrix satisfies a restrictive structural constraint---known as {\em incoherence}---on its row and column spaces. In these cases, the subset of elements is sampled uniformly at random. In this paper, we show that {\em any} rank-$ r $ $ n$-by-$ n $ matrix can be exactly recovered from as few as $O(nr \log^2 n)$ randomly chosen elements, provided this random choice is made according to a {\em specific biased distribution}: the probability of any element being sampled should be proportional to the sum of the leverage scores of the corresponding row, and column. Perhaps equally important, we show that this specific form of sampling is nearly necessary, in a natural precise sense; this implies that other perhaps more intuitive sampling schemes fail. We further establish three ways to use the above result for the setting when leverage scores are not known \textit{a priori}: (a) a sampling strategy for the case when only one of the row or column spaces are incoherent, (b) a two-phase sampling procedure for general matrices that first samples to estimate leverage scores followed by sampling for exact recovery, and (c) an analysis showing the advantages of weighted nuclear/trace-norm minimization over the vanilla un-weighted formulation for the case of non-uniform sampling.