The Comparative Study of Different Number of Particles in Clustering Based on Three-Layer Particle Swarm Optimization
To study how the different number of particles in clustering affect the performance of three-layer particle swarm optimization (THLPSO) that sets the global best location in each swarm to be the position of the particle in the swarm of the next layer, ten configurations of the different number of particles are compared. Fourteen benchmark functions, being in seven types with different circumstance, are used in the experiments. The experiments show that the searching ability of the algorithms is related to the number of particles in clustering, which is better with the number of particles transforming from as little as possible to as much as possible in each swarm when the function dimension is increasing from less to more. Finally, the original algorithm and THLPSO are compared to illustrate the efficiency of the proposed method.
KeywordsParticle swarm optimization hierarchy cluster
Unable to display preview. Download preview PDF.
- 2.Kennedy, J.: Stereotyping: Improving Particle Swarm Performance with Cluster Analysis. In: Proceedings of the 2000 Congress on Evolutionary Computation, vol. 2, pp. 1507–1512 (2000)Google Scholar