A New Clustering-Based Evolutionary Algorithm for Real- Valued Multimodal Optimization
Cătălin Stoean, Mike Preuß, Thomas Bartz–Beielstein, Ruxandra Stoean, Cătălin Stoean, Mike Preuß, Thomas Bartz–Beielstein, Ruxandra Stoean · 2009
Solving multimodal optimization tasks (problems with multiple global/local optimal solutions) by the state-of-the-art evolutionary algorithms (EAs) presumes separation of a population of individuals into subpopulations, each connected to a different optimum, with the aim of maintaining diversity for a longer period of time. Instead of using the typical separation that uses depends on a radius, present work proposes the employment of a clustering technique in order to distribute the candidate solutions to different species. Additionally, the proposed method corrects the separation by means of a mechanism that verifies the topological placement of the individuals in the fitness landscape with the purpose of connecting each species to a different optimum. The best individuals from each subpopulation are preserved from one generation to another in order to assure the conservation of the species. The method is applied on a set of benchmark functions that exhibit various properties, under multiple parameter settings, and the results demonstrate its great potential, especially of coping with relatively difficult problems under a limited budget of fitness evaluations.