English

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

Disordered Systems and Neural Networks 2022-03-22 v2 Machine Learning Statistics Theory Machine Learning Statistics Theory

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.

Keywords

Cite

@article{arxiv.2103.04902,
  title  = {Stochasticity helps to navigate rough landscapes: comparing gradient-descent-based algorithms in the phase retrieval problem},
  author = {Francesca Mignacco and Pierfrancesco Urbani and Lenka Zdeborová},
  journal= {arXiv preprint arXiv:2103.04902},
  year   = {2022}
}

Comments

28 pages, 11 figures

R2 v1 2026-06-23T23:53:05.111Z