Complexity Guarantees for Polyak Steps with Momentum

arXiv2002.00915MaRDI QIDQ6334058

Author name not available (Why is that?)

Publication date: 3 February 2020

Abstract: In smooth strongly convex optimization, knowledge of the strong convexity parameter is critical for obtaining simple methods with accelerated rates. In this work, we study a class of methods, based on Polyak steps, where this knowledge is substituted by that of the optimal value, $f_{*}$. We first show slightly improved convergence bounds than previously known for the classical case of simple gradient descent with Polyak steps, we then derive an accelerated gradient method with Polyak steps and momentum, along with convergence guarantees.

Has companion code repository: https://github.com/mathbarre/PerformanceEstimationPolyakSteps

This page was built for publication: Complexity Guarantees for Polyak Steps with Momentum

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6334058)