Backtracking strategies for accelerated descent methods with smooth composite objectives - Archive ouverte HAL Access content directly
Journal Articles SIAM Journal on Optimization Year : 2019

Backtracking strategies for accelerated descent methods with smooth composite objectives

Abstract

We present and analyse a backtracking strategy for a general Fast Iterative Shrink-age/Thresholding Algorithm which has been recently proposed in [11] for strongly convex objective functions. Differently from classical Armijo-type line searching, our backtracking rule allows for local increase and decrease of the Lipschitz constant estimate along the iterations. For such strategy accelerated convergence rates are proved and numerical results are shown for some exemplar imaging problems.
Fichier principal
Vignette du fichier
paper_FISTA_bt.pdf (710.43 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01596103 , version 1 (27-09-2017)

Identifiers

  • HAL Id : hal-01596103 , version 1

Cite

Luca Calatroni, Antonin Chambolle. Backtracking strategies for accelerated descent methods with smooth composite objectives. SIAM Journal on Optimization, 2019, 29 (3), pp.1772--1798. ⟨hal-01596103⟩
355 View
750 Download

Share

Gmail Facebook Twitter LinkedIn More