A beginner's guide to tuning methods

Elizabeth Montero, María Cristina Riff, Bertrand Neveu

Resultado de la investigación: Article

25 Citas (Scopus)

Resumen

Metaheuristic methods have been demonstrated to be efficient tools to solve hard optimization problems. Most metaheuristics define a set of parameters that must be tuned. A good setup of that parameter valuescan lead to take advantage of the metaheuristic capabilities to solve the problem at hand. Tuning strategiesare step by step methods based on multiple runs of the metaheuristic algorithm. In this study we comparefour automated tuning methods: F-Race, Revac, ParamILS and SPO. We evaluate the performance of eachmethod using a standard genetic algorithm for continuous function optimization. We discuss about therequirements of each method, the resources used and quality of solutions found in different scenarios.Finally we establish some guidelines that can help to choose the more appropriate tuning procedure.

Idioma originalEnglish
Páginas (desde-hasta)39-51
Número de páginas13
PublicaciónApplied Soft Computing Journal
Volumen17
DOI
EstadoPublished - 1 ene 2014

Huella dactilar

Tuning
Genetic algorithms

ASJC Scopus subject areas

  • Software

Citar esto

Montero, Elizabeth ; Riff, María Cristina ; Neveu, Bertrand. / A beginner's guide to tuning methods. En: Applied Soft Computing Journal. 2014 ; Vol. 17. pp. 39-51.
@article{eb01f89ae33446e49f151ef0f3fc1739,
title = "A beginner's guide to tuning methods",
abstract = "Metaheuristic methods have been demonstrated to be efficient tools to solve hard optimization problems. Most metaheuristics define a set of parameters that must be tuned. A good setup of that parameter valuescan lead to take advantage of the metaheuristic capabilities to solve the problem at hand. Tuning strategiesare step by step methods based on multiple runs of the metaheuristic algorithm. In this study we comparefour automated tuning methods: F-Race, Revac, ParamILS and SPO. We evaluate the performance of eachmethod using a standard genetic algorithm for continuous function optimization. We discuss about therequirements of each method, the resources used and quality of solutions found in different scenarios.Finally we establish some guidelines that can help to choose the more appropriate tuning procedure.",
keywords = "Evolutionary algorithms, Metaheuristics, Parameter setting problem, Tuning methods",
author = "Elizabeth Montero and Riff, {Mar{\'i}a Cristina} and Bertrand Neveu",
year = "2014",
month = "1",
day = "1",
doi = "10.1016/j.asoc.2013.12.017",
language = "English",
volume = "17",
pages = "39--51",
journal = "Applied Soft Computing",
issn = "1568-4946",
publisher = "Elsevier BV",

}

A beginner's guide to tuning methods. / Montero, Elizabeth; Riff, María Cristina; Neveu, Bertrand.

En: Applied Soft Computing Journal, Vol. 17, 01.01.2014, p. 39-51.

Resultado de la investigación: Article

TY - JOUR

T1 - A beginner's guide to tuning methods

AU - Montero, Elizabeth

AU - Riff, María Cristina

AU - Neveu, Bertrand

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Metaheuristic methods have been demonstrated to be efficient tools to solve hard optimization problems. Most metaheuristics define a set of parameters that must be tuned. A good setup of that parameter valuescan lead to take advantage of the metaheuristic capabilities to solve the problem at hand. Tuning strategiesare step by step methods based on multiple runs of the metaheuristic algorithm. In this study we comparefour automated tuning methods: F-Race, Revac, ParamILS and SPO. We evaluate the performance of eachmethod using a standard genetic algorithm for continuous function optimization. We discuss about therequirements of each method, the resources used and quality of solutions found in different scenarios.Finally we establish some guidelines that can help to choose the more appropriate tuning procedure.

AB - Metaheuristic methods have been demonstrated to be efficient tools to solve hard optimization problems. Most metaheuristics define a set of parameters that must be tuned. A good setup of that parameter valuescan lead to take advantage of the metaheuristic capabilities to solve the problem at hand. Tuning strategiesare step by step methods based on multiple runs of the metaheuristic algorithm. In this study we comparefour automated tuning methods: F-Race, Revac, ParamILS and SPO. We evaluate the performance of eachmethod using a standard genetic algorithm for continuous function optimization. We discuss about therequirements of each method, the resources used and quality of solutions found in different scenarios.Finally we establish some guidelines that can help to choose the more appropriate tuning procedure.

KW - Evolutionary algorithms

KW - Metaheuristics

KW - Parameter setting problem

KW - Tuning methods

UR - http://www.scopus.com/inward/record.url?scp=84893109228&partnerID=8YFLogxK

U2 - 10.1016/j.asoc.2013.12.017

DO - 10.1016/j.asoc.2013.12.017

M3 - Article

VL - 17

SP - 39

EP - 51

JO - Applied Soft Computing

JF - Applied Soft Computing

SN - 1568-4946

ER -