A Cooperative and Penalized Competitive Learning Approach to Gaussian Mixture Clustering

Competitive learning approaches with penalization or cooperation mechanism have been applied to unsupervised data clustering due to their attractive ability of automatic cluster number selection. In this paper, we further investigate the properties of different competitive strategies and propose a novel learning algorithm called Cooperative and Penalized Competitive Learning (CPCL), which implements the cooperation and penalization mechanisms simultaneously in a single competitive learning process. The integration of these two different kinds of competition mechanisms enables the CPCL to have good convergence speed, precision and robustness. Experiments on Gaussian mixture clustering are performed to investigate the proposed algorithm. The promising results demonstrate its superiority.