Topologias dinâmicas para modelo em ilhas usando evolução diferencial

Abstract

Over the last decades, the development of evolutionary metaheuristics has proportionated the increase the range of applications, as well as, the solution of complex problems. In particular, the Island Model (IM) is an important multi-population approach able to improve the performance of these metaheuristics in high complex problems. Despite its success, its performance is closely related to the appropriate choice of the IM parameters. One of those parameters is the migration topology. It is responsible by defining the migratory ow between the subpopulations of the model, impacting the quality of solutions and convergence time. As the majority of the topologies found in the literature are static and naive, a new concept of dynamic migratory ows, known as dynamic migrationtopology, has been studied. Several dynamic migration topologies for IM can be found in the literature. In order to understand the effects of these approaches on the performance of the IM using the Diffrential Evolution (DE) algorithm, this study presents a thorough experimental analysis of the existing approaches. Furthermore, an analysis of the topologies obtained by the studied approaches by means of complex networksmetrics also provides some insight on the interaction between their characteristics and the perfomance of the model. The results show that the dynamic migration topologies is able to improve the performance of the IM without significant loss in the final quality of solutions.