The reviewed record of science sign in
Pith

arxiv: 2103.04902 · v2 · pith:QVHHWI6S · submitted 2021-03-08 · cond-mat.dis-nn · cs.LG· math.ST· stat.ML· stat.TH

Stochasticity helps to navigate rough landscapes: comparing gradient-descent-based algorithms in the phase retrieval problem

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:QVHHWI6Srecord.jsonopen to challenge →

classification cond-mat.dis-nn cs.LGmath.STstat.MLstat.TH
keywords descentgradientalgorithmsgeneralizationphaseretrievalablealgorithm
0
0 comments X
read the original abstract

In this paper we investigate how gradient-based algorithms such as gradient descent, (multi-pass) stochastic gradient descent, its persistent variant, and the Langevin algorithm navigate non-convex loss-landscapes and which of them is able to reach the best generalization error at limited sample complexity. We consider the loss landscape of the high-dimensional phase retrieval problem as a prototypical highly non-convex example. We observe that for phase retrieval the stochastic variants of gradient descent are able to reach perfect generalization for regions of control parameters where the gradient descent algorithm is not. We apply dynamical mean-field theory from statistical physics to characterize analytically the full trajectories of these algorithms in their continuous-time limit, with a warm start, and for large system sizes. We further unveil several intriguing properties of the landscape and the algorithms such as that the gradient descent can obtain better generalization properties from less informed initializations.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.