arXiv Analytics

Sign in

arXiv:1905.07436 [math.OC]AbstractReferencesReviewsResources

A Dynamical Systems Perspective on Nesterov Acceleration

Michael Muehlebach, Michael I. Jordan

Published 2019-05-17Version 1

We present a dynamical system framework for understanding Nesterov's accelerated gradient method. In contrast to earlier work, our derivation does not rely on a vanishing step size argument. We show that Nesterov acceleration arises from discretizing an ordinary differential equation with a semi-implicit Euler integration scheme. We analyze both the underlying differential equation as well as the discretization to obtain insights into the phenomenon of acceleration. The analysis suggests that a curvature-dependent damping term lies at the heart of the phenomenon. We further establish connections between the discretized and the continuous-time dynamics.

Comments: 11 pages, 4 figures, to appear in the Proceedings of the 36th International Conference on Machine Learning
Categories: math.OC, cs.LG, cs.SY, stat.ML
Related articles: Most relevant | Search more
arXiv:2305.08536 [math.OC] (Published 2023-05-15)
A Dynamical Systems Perspective on Discrete Optimization
arXiv:2004.00424 [math.OC] (Published 2020-03-29)
Solving the inverse problem for an ordinary differential equation using conjugation
Alfaro Vigo et al.
arXiv:1909.07492 [math.OC] (Published 2019-09-16)
On-line Non-Convex Constrained Optimization