19
4

Implicit regularization and solution uniqueness in over-parameterized matrix sensing

Kelly Geyer
Anastasios Kyrillidis
A. Kalev
Abstract

We consider whether algorithmic choices in over-parameterized linear matrix factorization introduce implicit regularization. We focus on noiseless matrix sensing over rank-rr positive semi-definite (PSD) matrices in Rn×n\mathbb{R}^{n \times n}, with a sensing mechanism that satisfies restricted isometry properties (RIP). The algorithm we study is \emph{factored gradient descent}, where we model the low-rankness and PSD constraints with the factorization UUUU^\top, for URn×rU \in \mathbb{R}^{n \times r}. Surprisingly, recent work argues that the choice of rnr \leq n is not pivotal: even setting URn×nU \in \mathbb{R}^{n \times n} is sufficient for factored gradient descent to find the rank-rr solution, which suggests that operating over the factors leads to an implicit regularization. In this contribution, we provide a different perspective to the problem of implicit regularization. We show that under certain conditions, the PSD constraint by itself is sufficient to lead to a unique rank-rr matrix recovery, without implicit or explicit low-rank regularization. \emph{I.e.}, under assumptions, the set of PSD matrices, that are consistent with the observed data, is a singleton, regardless of the algorithm used.

View on arXiv
Comments on this paper