Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization

Synchronous and asynchronous algorithms are presented for distributed minimax optimization. The objective here is to realize the minimization of the maximum of component functions over the standard multi-agent network, where each node of the network knows its own function and it exchanges its decisi...

Full description

Bibliographic Details
Main Authors: Kenta Hanada, Ryosuke Morita, Takayuki Wada, Yasumasa Fujisaki
Format: Article
Language:English
Published: Taylor & Francis Group 2017-11-01
Series:SICE Journal of Control, Measurement, and System Integration
Subjects:
Online Access:http://dx.doi.org/10.9746/jcmsi.10.557
_version_ 1797661055288082432
author Kenta Hanada
Ryosuke Morita
Takayuki Wada
Yasumasa Fujisaki
author_facet Kenta Hanada
Ryosuke Morita
Takayuki Wada
Yasumasa Fujisaki
author_sort Kenta Hanada
collection DOAJ
description Synchronous and asynchronous algorithms are presented for distributed minimax optimization. The objective here is to realize the minimization of the maximum of component functions over the standard multi-agent network, where each node of the network knows its own function and it exchanges its decision variable with its neighbors. In fact, the proposed algorithms are standard consensus and gossip based subgradient methods, while the original minimax optimization is recast as minimization of the sum of component functions by using a p-norm approximation. A scalable step size depending on the approximation ratio p is also presented in order to avoid slow convergence. Numerical examples illustrate that the algorithms with this step size work well even in the high approximation ratios.
first_indexed 2024-03-11T18:38:44Z
format Article
id doaj.art-5eb1c2ca739f417983bd907c9d0c5d91
institution Directory Open Access Journal
issn 1884-9970
language English
last_indexed 2024-03-11T18:38:44Z
publishDate 2017-11-01
publisher Taylor & Francis Group
record_format Article
series SICE Journal of Control, Measurement, and System Integration
spelling doaj.art-5eb1c2ca739f417983bd907c9d0c5d912023-10-12T13:43:54ZengTaylor & Francis GroupSICE Journal of Control, Measurement, and System Integration1884-99702017-11-0110655756210.9746/jcmsi.10.55712103177Simple Synchronous and Asynchronous Algorithms for Distributed Minimax OptimizationKenta Hanada0Ryosuke Morita1Takayuki Wada2Yasumasa Fujisaki3Graduate School of Information Science and Technology, Osaka UniversityFaculty of Engineering, Gifu UniversityGraduate School of Information Science and Technology, Osaka UniversityGraduate School of Information Science and Technology, Osaka UniversitySynchronous and asynchronous algorithms are presented for distributed minimax optimization. The objective here is to realize the minimization of the maximum of component functions over the standard multi-agent network, where each node of the network knows its own function and it exchanges its decision variable with its neighbors. In fact, the proposed algorithms are standard consensus and gossip based subgradient methods, while the original minimax optimization is recast as minimization of the sum of component functions by using a p-norm approximation. A scalable step size depending on the approximation ratio p is also presented in order to avoid slow convergence. Numerical examples illustrate that the algorithms with this step size work well even in the high approximation ratios.http://dx.doi.org/10.9746/jcmsi.10.557distributed algorithmsminimax optimizationsynchronous algorithmsasynchronous algorithms
spellingShingle Kenta Hanada
Ryosuke Morita
Takayuki Wada
Yasumasa Fujisaki
Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization
SICE Journal of Control, Measurement, and System Integration
distributed algorithms
minimax optimization
synchronous algorithms
asynchronous algorithms
title Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization
title_full Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization
title_fullStr Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization
title_full_unstemmed Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization
title_short Simple Synchronous and Asynchronous Algorithms for Distributed Minimax Optimization
title_sort simple synchronous and asynchronous algorithms for distributed minimax optimization
topic distributed algorithms
minimax optimization
synchronous algorithms
asynchronous algorithms
url http://dx.doi.org/10.9746/jcmsi.10.557
work_keys_str_mv AT kentahanada simplesynchronousandasynchronousalgorithmsfordistributedminimaxoptimization
AT ryosukemorita simplesynchronousandasynchronousalgorithmsfordistributedminimaxoptimization
AT takayukiwada simplesynchronousandasynchronousalgorithmsfordistributedminimaxoptimization
AT yasumasafujisaki simplesynchronousandasynchronousalgorithmsfordistributedminimaxoptimization