PMAFC: A New Probabilistic Memetic Algorithm Based Fuzzy Clustering

In this article, a new stochastic approach in form of memetic algorithm for fuzzy clustering is presented. The proposed probabilistic memetic algorithm based fuzzy clustering technique uses real-coded encoding of the cluster centres and two fuzzy clustering validity measures to compute a priori probability for an objective function. Moreover, the adaptive arithmetic recombination and opposite based local search techniques are used to get better performance of the proposed algorithm by exploring the search space more powerfully. The performance of the proposed clustering algorithm has been compared with that of some well-known existing clustering algorithms for four synthetic and two real life data sets. Statistical significance test based on analysis of variance (ANOVA) has been conducted to establish the statistical significance of the superior performance of the proposed clustering algorithm. Matlab version of the software is available at http://sysbio.icm.edu.pl/memetic.