Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter?

Elizabeth Montero, María Cristina Riff, Leslie Pérez-Caceres, Carlos A. Coello Coello

Resultado de la investigación: Conference contribution

5 Citas (Scopus)

Resumen

Currently, there exist several offline calibration techniques that can be used to fine-tune the parameters of a metaheuristic. Such techniques require, however, to perform a considerable number of independent runs of the metaheuristic in order to obtain meaningful information. Here, we are interested on the use of this information for assisting the algorithm designer to discard components of a metaheuristic (e.g., an evolutionary operator) that do not contribute to improving its performance (we call them "ineffective components"). In our study, we experimentally analyze the information obtained from three offline calibration techniques: F-Race, ParamILS and Revac. Our preliminary results indicate that these three calibration techniques provide different types of information, which makes it necessary to conduct a more in-depth analysis of the data obtained, in order to detect the ineffective components that are of our interest.

Idioma originalEnglish
Título de la publicación alojadaParallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings
Páginas306-315
Número de páginas10
EdiciónPART 1
DOI
EstadoPublished - 24 sep 2012
Evento12th International Conference on Parallel Problem Solving from Nature, PPSN 2012 - Taormina, Italy
Duración: 1 sep 20125 sep 2012

Serie de la publicación

NombreLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NúmeroPART 1
Volumen7491 LNCS
ISSN (versión impresa)0302-9743
ISSN (versión digital)1611-3349

Conference

Conference12th International Conference on Parallel Problem Solving from Nature, PPSN 2012
PaísItaly
CiudadTaormina
Período1/09/125/09/12

Huella dactilar

Tuning
Metaheuristics
Calibration
Necessary
Operator

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Citar esto

Montero, E., Riff, M. C., Pérez-Caceres, L., & Coello Coello, C. A. (2012). Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter? En Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings (PART 1 ed., pp. 306-315). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7491 LNCS, N.º PART 1). https://doi.org/10.1007/978-3-642-32937-1_31
Montero, Elizabeth ; Riff, María Cristina ; Pérez-Caceres, Leslie ; Coello Coello, Carlos A. / Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter?. Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings. PART 1. ed. 2012. pp. 306-315 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1).
@inproceedings{3d6d9698d7bf47a49662973ec121fe2a,
title = "Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter?",
abstract = "Currently, there exist several offline calibration techniques that can be used to fine-tune the parameters of a metaheuristic. Such techniques require, however, to perform a considerable number of independent runs of the metaheuristic in order to obtain meaningful information. Here, we are interested on the use of this information for assisting the algorithm designer to discard components of a metaheuristic (e.g., an evolutionary operator) that do not contribute to improving its performance (we call them {"}ineffective components{"}). In our study, we experimentally analyze the information obtained from three offline calibration techniques: F-Race, ParamILS and Revac. Our preliminary results indicate that these three calibration techniques provide different types of information, which makes it necessary to conduct a more in-depth analysis of the data obtained, in order to detect the ineffective components that are of our interest.",
keywords = "algorithm design process, fine-tuning methods, ineffective operators",
author = "Elizabeth Montero and Riff, {Mar{\'i}a Cristina} and Leslie P{\'e}rez-Caceres and {Coello Coello}, {Carlos A.}",
year = "2012",
month = "9",
day = "24",
doi = "10.1007/978-3-642-32937-1_31",
language = "English",
isbn = "9783642329364",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 1",
pages = "306--315",
booktitle = "Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings",
edition = "PART 1",

}

Montero, E, Riff, MC, Pérez-Caceres, L & Coello Coello, CA 2012, Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter? En Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings. PART 1 ed., Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), n.º PART 1, vol. 7491 LNCS, pp. 306-315, 12th International Conference on Parallel Problem Solving from Nature, PPSN 2012, Taormina, Italy, 1/09/12. https://doi.org/10.1007/978-3-642-32937-1_31

Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter? / Montero, Elizabeth; Riff, María Cristina; Pérez-Caceres, Leslie; Coello Coello, Carlos A.

Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings. PART 1. ed. 2012. p. 306-315 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7491 LNCS, N.º PART 1).

Resultado de la investigación: Conference contribution

TY - GEN

T1 - Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter?

AU - Montero, Elizabeth

AU - Riff, María Cristina

AU - Pérez-Caceres, Leslie

AU - Coello Coello, Carlos A.

PY - 2012/9/24

Y1 - 2012/9/24

N2 - Currently, there exist several offline calibration techniques that can be used to fine-tune the parameters of a metaheuristic. Such techniques require, however, to perform a considerable number of independent runs of the metaheuristic in order to obtain meaningful information. Here, we are interested on the use of this information for assisting the algorithm designer to discard components of a metaheuristic (e.g., an evolutionary operator) that do not contribute to improving its performance (we call them "ineffective components"). In our study, we experimentally analyze the information obtained from three offline calibration techniques: F-Race, ParamILS and Revac. Our preliminary results indicate that these three calibration techniques provide different types of information, which makes it necessary to conduct a more in-depth analysis of the data obtained, in order to detect the ineffective components that are of our interest.

AB - Currently, there exist several offline calibration techniques that can be used to fine-tune the parameters of a metaheuristic. Such techniques require, however, to perform a considerable number of independent runs of the metaheuristic in order to obtain meaningful information. Here, we are interested on the use of this information for assisting the algorithm designer to discard components of a metaheuristic (e.g., an evolutionary operator) that do not contribute to improving its performance (we call them "ineffective components"). In our study, we experimentally analyze the information obtained from three offline calibration techniques: F-Race, ParamILS and Revac. Our preliminary results indicate that these three calibration techniques provide different types of information, which makes it necessary to conduct a more in-depth analysis of the data obtained, in order to detect the ineffective components that are of our interest.

KW - algorithm design process

KW - fine-tuning methods

KW - ineffective operators

UR - http://www.scopus.com/inward/record.url?scp=84866378792&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-32937-1_31

DO - 10.1007/978-3-642-32937-1_31

M3 - Conference contribution

SN - 9783642329364

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 306

EP - 315

BT - Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings

ER -

Montero E, Riff MC, Pérez-Caceres L, Coello Coello CA. Are state-of-the-art fine-tuning algorithms able to detect a dummy parameter? En Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings. PART 1 ed. 2012. p. 306-315. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 1). https://doi.org/10.1007/978-3-642-32937-1_31