Non-Convex Rank/Sparsity Regularization and Local Minima
Carl Olsson, Marcus Carlsson, Fredrik Andersson, Viktor Larsson
Abstract
This paper considers the problem of recovering either a low rank matrix or a sparse vector from observations of linear combinations of the vector or matrix elements. Recent methods replace the non-convex regularization with l1 or nuclear norm relaxations. It is well known that this approach recovers near optimal solutions if a so called restricted isometry property (RIP) holds. On the other hand it also has a shrinking bias which can degrade the solution. In this paper we study an alternative non-convex regularization term that does not suffer from this bias. Our main theoretical results show that if a RIP holds then the stationary points are often well separated, in the sense that their differences must be of high cardinality/rank. Thus, with a suitable initial solution the approach is unlikely to fall into a bad local minimum. Our numerical tests show that the approach is likely to converge to a better solution than standard l1/nuclear-norm relaxation even when starting from trivial initializations. In many cases our results can also be used to verify global optimality of our method.
BibTeX
@inproceedings{iccv2017_nonconvexrankspa,
title = {Non-Convex Rank/Sparsity Regularization and Local Minima},
author = {Carl Olsson and Marcus Carlsson and Fredrik Andersson and Viktor Larsson},
booktitle = {ICCV 2017},
year = {2017}
}