Descripción
|
|
---|---|
One of the most promising areas in which probabilisticgraphical models have shown an incipient activity is thefield of heuristic optimization and, in particular, in the Estimationof Distribution Algorithms (EDAs). EDAs constitutea well-known family of Evolutionary Computationtechniques, similar to Genetic Algorithms. Due to their inherentparallelism, different research lines have been studiedtrying to improve EDAs from the point of view of executiontime and/or accuracy. Among these proposals, wefocus on the so-called island-based models. This approachdefines several islands (EDA instances) running independentlyand exchanging information with a given frequency.The information sent by the islands can be a set of individualsor a probabilistic model. This paper presents a comparativestudy of both information exchanging techniquesfor a univariate EDA (UMDAg) over a wide set of parametersand problems ¿the standard benchmark developed forthe IEEEWorkshop on Evolutionary Algorithms and otherMetaheuristics for Continuous Optimization Problems ofthe ISDA 2009 Conference. The study concludes that theconfigurations based on migrating individuals obtain betterresults. | |
Internacional
|
No |
Nombre congreso
|
Proceedings of the Tenth IASTED International Conference on Artificial Intelligence and Applications (2010) |
Tipo de participación
|
960 |
Lugar del congreso
|
|
Revisores
|
Si |
ISBN o ISSN
|
978-0-88986-818-2 |
DOI
|
|
Fecha inicio congreso
|
15/02/2010 |
Fecha fin congreso
|
17/02/2010 |
Desde la página
|
255 |
Hasta la página
|
262 |
Título de las actas
|
Migrating Individuals and Probabilistic Models on DEDAS: a Comparison on Continuous Functions |