A parallel genetic algorithm for clustering

Parallelization of genetic algorithms (GAs) has received considerable attention in recent years. Reasons for this are the availability of suitable computational resources and the need for solving harder problems in reasonable time. We describe a new parallel self-adaptive GA for solving the data clustering problem. The algorithm utilizes island parallelization using a genebank model, in which GA processes communicate with each other through the genebank process. This model allows one to implement different migration topologies in an easy manner. Experiments show that significant speedup is reached by parallelization. The effect of migration parameters is also studied and the development of population diversity is examined by several measures, some of which are new.