Small World Particle Swarm Optimizer for Data Clustering

Abstract

Particle swarm is a stochastic optimization paradigm inspired by the concepts of social psychology and artificial intelligence. Population topology plays significant role in the performance of PSO. It determines the way in which particles communicate and share information with each other. Topology can be depicted as a network model. Regular networks are highly clustered but the characteristic path length grows linearly with the increase in number of vertices. On the contrary, random networks are not highly clustered but they have small characteristic path length. Small world networks have a distinctive combination of regular and random networks i.e. highly clustered and small characteristic path length. This paper presents a novel algorithm for data clustering by incorporating the concept of small world in particle swarm optimization. Efficiency of the proposed methodology is tested by applying it on five standard benchmark data set. Results obtained are compared with another PSO variant. Comparative study demonstrates the effectiveness of the proposed approach.