A Stochastic Disturbance of Particle Swarm Optimization for K-Means Clustering Method
This paper presents a hybrid-clustering algorithm that is a stochastic disturbance of particle swarm optimization (PSO) for K-means clustering method (SDPSO-K). The proposed algorithm can improve the particle global searching ability in PSO to avoid the K-means disadvantage of being easily trapped in a local optimal solution and to save the expensive computational cost of PSO clustering. The performance of the SDPSO-K, compared with three recently developed modified PSO techniques and related clustering algorithms for six datasets, indicates that the SDPSO-K algorithm is clearly and consistently superior in terms of precision and robustness.
J. Y. Chen "A Stochastic Disturbance of Particle Swarm Optimization for K-Means Clustering Method", Advanced Materials Research, Vols. 268-270, pp. 10-15, 2011