Cargando…

An incremental mirror descent subgradient algorithm with random sweeping and proximal step

We investigate the convergence properties of incremental mirror descent type subgradient algorithms for minimizing the sum of convex functions. In each step, we only evaluate the subgradient of a single component function and mirror it back to the feasible domain, which makes iterations very cheap t...

Descripción completa

Detalles Bibliográficos
Autores principales: Boţ, Radu Ioan, Böhm, Axel
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Taylor & Francis 2018
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6382287/
https://www.ncbi.nlm.nih.gov/pubmed/30828224
http://dx.doi.org/10.1080/02331934.2018.1482491
_version_ 1783396642981937152
author Boţ, Radu Ioan
Böhm, Axel
author_facet Boţ, Radu Ioan
Böhm, Axel
author_sort Boţ, Radu Ioan
collection PubMed
description We investigate the convergence properties of incremental mirror descent type subgradient algorithms for minimizing the sum of convex functions. In each step, we only evaluate the subgradient of a single component function and mirror it back to the feasible domain, which makes iterations very cheap to compute. The analysis is made for a randomized selection of the component functions, which yields the deterministic algorithm as a special case. Under supplementary differentiability assumptions on the function which induces the mirror map, we are also able to deal with the presence of another term in the objective function, which is evaluated via a proximal type step. In both cases, we derive convergence rates of [Image: see text] in expectation for the kth best objective function value and illustrate our theoretical findings by numerical experiments in positron emission tomography and machine learning.
format Online
Article
Text
id pubmed-6382287
institution National Center for Biotechnology Information
language English
publishDate 2018
publisher Taylor & Francis
record_format MEDLINE/PubMed
spelling pubmed-63822872019-03-01 An incremental mirror descent subgradient algorithm with random sweeping and proximal step Boţ, Radu Ioan Böhm, Axel Optimization Article We investigate the convergence properties of incremental mirror descent type subgradient algorithms for minimizing the sum of convex functions. In each step, we only evaluate the subgradient of a single component function and mirror it back to the feasible domain, which makes iterations very cheap to compute. The analysis is made for a randomized selection of the component functions, which yields the deterministic algorithm as a special case. Under supplementary differentiability assumptions on the function which induces the mirror map, we are also able to deal with the presence of another term in the objective function, which is evaluated via a proximal type step. In both cases, we derive convergence rates of [Image: see text] in expectation for the kth best objective function value and illustrate our theoretical findings by numerical experiments in positron emission tomography and machine learning. Taylor & Francis 2018-06-14 /pmc/articles/PMC6382287/ /pubmed/30828224 http://dx.doi.org/10.1080/02331934.2018.1482491 Text en © 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group http://creativecommons.org/licenses/by/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
spellingShingle Article
Boţ, Radu Ioan
Böhm, Axel
An incremental mirror descent subgradient algorithm with random sweeping and proximal step
title An incremental mirror descent subgradient algorithm with random sweeping and proximal step
title_full An incremental mirror descent subgradient algorithm with random sweeping and proximal step
title_fullStr An incremental mirror descent subgradient algorithm with random sweeping and proximal step
title_full_unstemmed An incremental mirror descent subgradient algorithm with random sweeping and proximal step
title_short An incremental mirror descent subgradient algorithm with random sweeping and proximal step
title_sort incremental mirror descent subgradient algorithm with random sweeping and proximal step
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6382287/
https://www.ncbi.nlm.nih.gov/pubmed/30828224
http://dx.doi.org/10.1080/02331934.2018.1482491
work_keys_str_mv AT botraduioan anincrementalmirrordescentsubgradientalgorithmwithrandomsweepingandproximalstep
AT bohmaxel anincrementalmirrordescentsubgradientalgorithmwithrandomsweepingandproximalstep
AT botraduioan incrementalmirrordescentsubgradientalgorithmwithrandomsweepingandproximalstep
AT bohmaxel incrementalmirrordescentsubgradientalgorithmwithrandomsweepingandproximalstep