Harmonic Mean Iteratively Reweighted Least Squares for Low-Rank Matrix Recovery

arXiv1703.05038MaRDI QIDQ6284346FDOQ6284346

Authors: Christian Kümmerle, Juliane Sigl 

Publication date: 15 March 2017

Abstract: We propose a new iteratively reweighted least squares (IRLS) algorithm for the recovery of a matrix $Xinmathbb{C}^{d_{1}imes{d}_2}$ of rank $rllmin(d_1,d_2)$ from incomplete linear observations, solving a sequence of low complexity linear problems. The easily implementable algorithm, which we call harmonic mean iteratively reweighted least squares (HM-IRLS), optimizes a non-convex Schatten-$p$ quasi-norm penalization to promote low-rankness and carries three major strengths, in particular for the matrix completion setting. First, we observe a remarkable global convergence behavior of the algorithm's iterates to the low-rank matrix for relevant, interesting cases, for which any other state-of-the-art optimization approach fails the recovery. Secondly, HM-IRLS exhibits an empirical recovery probability close to $1$ even for a number of measurements very close to the theoretical lower bound $r(d_1+d_2-r)$, i.e., already for significantly fewer linear observations than any other tractable approach in the literature. Thirdly, HM-IRLS exhibits a locally superlinear rate of convergence (of order $2-p$) if the linear observations fulfill a suitable null space property. While for the first two properties we have so far only strong empirical evidence, we prove the third property as our main theoretical result.

Mathematics Subject Classification ID

Estimation in multivariate analysis (62H12) Nonconvex programming, global optimization (90C26)

This page was built for publication: Harmonic Mean Iteratively Reweighted Least Squares for Low-Rank Matrix Recovery

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6284346)