Multi Swarm Based Ensemble Clustering

Abstract

This research paper is mainly targeted towards the Clustering Approach using the proposed  Genetic Algorithm called “Multi Swarm Based Ensemble Clustering Algorithm”. This clustering algorithm is inspired by Eberhart and Kennedy’s Particle Swarm Optimization (PSO) [1] and Cohen and de Castro Particle Swarm Clustering (PSC) [2] and Yuwono’s Rapid Centroid Estimation (RCE) . The proposed algorithm improves PSO and PSC in terms of memory and computational efficiency, capability to automatically determine the number of clusters, and gracefully handle non-convex datasets in quasilinear complexity. Benefiting from the robustness of swarm intelligence, the versatility of voronoi tessellation and the flexibility of graph algorithms, the proposed algorithm is designed to discover natural groupings in both convex and non-convex data.

Key Words: Particle Swarm Clustering; Consensus Clustering;  Multi Swarm based PSO; Data Clustering.