The parallel multi-objective genetic algorithm in fuzzy clustering for discrete elements

The paper proposes a multi-objective genetic algorithm in fuzzy clustering for discrete elements (PGC). The proposed algorithm is built for the fuzzy clustering problem using two objective functions. The first objective function is based on the DB index which used to evaluate separation of inside and outside of clusters. The second function utilized to determine the fuzzy relationship between each element and central clusters. In addition, author also proposes some improvement for the crossover, mutation and selection operators, and apply the parallel technique for the genetic algorithm. These improvements decreased the computing time and cost of PGC. In addition, PGC is also illustrated step by step via simulated data set, it also shows advantageous while compare to other algorithms including k-means, fuzzy c-means and Chen’s algorithm. This algorithm is built based on the Matlab procedures. Keywords. Parallel genetic algorithm, unsupervised learning, image classification, fuzzy clustering.