Optimizing Parameters of Fuzzy c-Means Clustering Algorithm

Abstract

For overcoming the shortcoming that Fuzzy c-Means (FCM) clustering algorithm seriously depends on the initial values of clustering numbers (c) and fuzzy exponent (m), we introduce genetic algorithm to find the pair parameters of FCM simultaneity. In the proposed algorithm, the clustering numbers and the fuzzy exponent are controlled by a binary code. In order to optimize the two parameters, new methods to code, decode, crossover and establish fitness function have been proposed. Results demonstrating the superiority of the proposed method, as compared to other method that only use validity index to find the clustering numbers (c), are provided for several real-life and artificial data sets.