Method of fuzzy clustering with genetic algorithm

Abstract

the original method of fuzzy clustering using genetic algorithm is proposed. The chromosomes of the population of genetic algorithm have variable length and encode the possible partition of the data. The method enables in one GA run to find the near optimal data partition into clusters, to determine the cluster number and simultaneously to speed up clustering process comparing to other known methods. Classification rules, constructed on the basis of clustering results, obtained by proposed method, provide more accurate classification in comparison with results of classical FCM-method.