C Conferentia Proceedings
CILAMCE2015-0147 COMPUTATIONAL INTELLIGENCE TECHNIQUES FOR OPTIMIZATION AND DATA MODELING

Migration policies to improve exploration in parallel island models for optimization via metaheuristics

Thiago Tavares Magalhães1; Eduardo Krempser2; Helio José Corrêa Barbosa1

1 Laboratório Nacional de Computação Científica (LNCC); 2 Fundação Oswaldo Cruz (FIOCRUZ)

doi:10.20906/CPS/CILAMCE2015-0147

Resumo

Island models are an interesting and powerful alternative in order to achieve at least two objectives in optimization: to increase the speed up of the algorithms (taking advantage of the growing computational power) and, at the same time, to improve the quality of the obtained solutions. Among many others, the "migration policy" is an important aspect that guide the behavior of the model. We suggested an useful topology of communication between the islands that compose the model, and employed 10 different policies, among well known and new suggestions of implementations. Were performed comparisons, applying a hybrid AG+DE model to solve a benchmark including 30 scalable problems with different levels of complexity, for 30 dimensions. The quality of the obtained results by the different techniques was studied, as well the distinct behaviors, in terms of sub populations diversity.

Palavras-chave: Meta heuristics; Parallel optimization; Migration policy

Como citar

Thiago Tavares Magalhães; Eduardo Krempser; Helio José Corrêa Barbosa. “Migration policies to improve exploration in parallel island models for optimization via metaheuristics”. XXXVI Ibero-Latin American Congress on Computational Methods in Engineering. CILAMCE2015. 2015. DOI: 10.20906/CPS/CILAMCE2015-0147