Intelligent hybrid approach for data clustering

Clustering is unsupervised learning method to extract hidden patterns and disciplines. Swarm intelligence deals with natural and artificial systems composed of many individuals that coordinate using decentralized control and self-organization. In this paper, we propose a new Swarm Intelligence based hybrid method for data clustering. The main difficulty in clustering is the numbers of resulting clusters are unknown and another difficulty is that the clustering algorithms falls into local optima. The Swarm Intelligence algorithm, Particle Swarm Optimization (PSO) and Bee Algorithm (BA) performs local and global search simultaneously. The proposed method based on PSO-BA for data clustering improves the accuracy of clustering and overcome the difficulty of local optima. This algorithm has been tested on four well-known real datasets and compared with other popular heuristics algorithm in clustering, such as K-means, Genetic Algorithm, Bee Algorithm and Particle Swarm Optimization algorithm and got promising results.