An Efficient Clustering Approach utilizing an Advanced Particle Swarm Optimization Variant

Abstract

Clustering plays an important role in almost every field of real time applications and whenever optimization is the core concern, good quality clustering is crucial. Different optimization techniques are proposed with respect to clustering and find its applications in areas like Image Processing and Pattern Recognition, Finance, Communication Networks, Biological Sequences, etc. This paper uses a superior variant of Particle Swarm Optimization (PSO) algorithm based on one of the density based clustering methodologies i.e. Subtractive Clustering (SC). The implementation of the algorithm proved its excellence in numerical and text data clustering by addressing the several issues those were came across in the literature survey of PSO based clustering techniques. The algorithm is intended to cluster image datasets for fine classification of images. The performance of the algorithm is evaluated against K-means, K-PSO, Subtractive-PSO over synthesized and MRI image datasets with respect to quantization error, intra and inter cluster distances. The obtained results showed a better or comparable performance.