Adaptive Particle Swarm Optimization Employing Fuzzy Logic

Swarm Intelligence algorithms, in many optimization problems, have constantly served a purpose of global search method. One of the problems confronted during optimization is clustering problem. Input for a clustering process is a set of data which are then organized into a number of sub-groups. Modern studies have recommended that partitioned or segregated clustering algorithms are more appropriate for clustering of wide and huge datasets. One of the most frequent partitional clustering algorithms is K-Means. K-means algorithm shows a more rapid convergence than PSO but then against local optimal area is generally trapped depending on the random values of initial centroids. An efficient hybrid method is presented in this paper, namely particle swarm optimization with fuzzy logic or adaptive particle swarm optimization (APSO) to resolve data clustering problem. The PSO algorithm does find a good or near optimal solution in reasonable time, but its presentation was enhanced by seeding the initial swarm with fuzzifier function. The adaptive fuzzy particle swarm optimization algorithm (APSO) is compared with k-means using total execution time and clustering group error. It is discovered that the total execution time for APSO method outperforms the k-means and had higher solution quality in terms of clustering group error.