Combining K-Means and K-Harmonic with Fish School Search Algorithm for data clustering task on graphics processing units

Abstract

Data clustering is related to the split of a set of objects into smaller groups with common features. Several optimization techniques have been proposed to increase the performance of clustering algorithms. Swarm Intelligence (SI) algorithms are concerned with optimization problems and they have been successfully applied to different domains. In this work, a Swarm Clustering Algorithm (SCA) is proposed based on the standard K-Means and on K-Harmonic Means (KHM) clustering algorithms, which are used as fitness functions for a SI algorithm: Fish School Search (FSS). The motivation is to exploit the search capability of SI algorithms and to avoid the major limitation of falling into locally optimal values of the K-Means algorithm. Because of the inherent parallel nature of the SI algorithms, since the fitness function can be evaluated for each individual in an isolated manner, we have developed the parallel implementation on GPU of the SCAs, comparing the performances with their serial implementation. The interest behind proposing SCA is to verify the ability of FSS algorithm to deal with the clustering task and to study the difference of performance of FSS-SCA implemented on CPU and on GPU. Experiments with 13 benchmark datasets have shown similar or slightly better quality of the results compared to standard K-Means algorithm and Particle Swarm Algorithm (PSO) algorithm. There results of using FSS for clustering are promising.