Accelerated Bregman proximal gradient methods for relatively smooth convex optimization

Embargo End Date
2022-04-07

Type
Article

Authors
Hanzely, Filip
Richtarik, Peter
Xiao, Lin

KAUST Department
Applied Mathematics and Computational Science Program
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
Computer Science Program

Preprint Posting Date
2018-08-09

Online Publication Date
2021-04-07

Print Publication Date
2021-06

Date
2021-04-07

Submitted Date
2020-04-24

Abstract
We consider the problem of minimizing the sum of two convex functions: one is differentiable and relatively smooth with respect to a reference convex function, and the other can be nondifferentiable but simple to optimize. We investigate a triangle scaling property of the Bregman distance generated by the reference convex function and present accelerated Bregman proximal gradient (ABPG) methods that attain an O(k-γ) convergence rate, where γ∈ (0 , 2] is the triangle scaling exponent (TSE) of the Bregman distance. For the Euclidean distance, we have γ= 2 and recover the convergence rate of Nesterov’s accelerated gradient methods. For non-Euclidean Bregman distances, the TSE can be much smaller (say γ≤ 1), but we show that a relaxed definition of intrinsic TSE is always equal to 2. We exploit the intrinsic TSE to develop adaptive ABPG methods that converge much faster in practice. Although theoretical guarantees on a fast convergence rate seem to be out of reach in general, our methods obtain empirical O(k- 2) rates in numerical experiments on several applications and provide posterior numerical certificates for the fast rates.

Citation
Hanzely, F., Richtárik, P., & Xiao, L. (2021). Accelerated Bregman proximal gradient methods for relatively smooth convex optimization. Computational Optimization and Applications. doi:10.1007/s10589-021-00273-8

Acknowledgements
We thank Haihao Lu, Robert Freund and Yurii Nesterov for helpful conversations. We are also grateful to the anonymous referees whose comments helped improving the clarity of the paper. Peter Richtárik acknowledges the support of the KAUST Baseline Research Funding Scheme.

Publisher
Springer Science and Business Media LLC

Journal
Computational Optimization and Applications

DOI
10.1007/s10589-021-00273-8

arXiv
1808.03045

Additional Links
http://link.springer.com/10.1007/s10589-021-00273-8

Permanent link to this record

Version History

Now showing 1 - 2 of 2
VersionDateSummary
2*
2021-04-21 07:54:21
Published in journal
2019-05-29 07:28:01
* Selected version