A cooperative and penalized competitive learning approach to Gaussian mixture clustering
ICANN'10 Proceedings of the 20th international conference on Artificial neural networks: Part III
| Hi-index | 0.00 |
Rival Penalized Competitive Learning (RPCL) and its variants can perform clustering analysis efficiently with the ability of selecting the cluster number automatically. Although they have been widely applied in a variety of research areas, some of their problems have not yet been solved. Based on the semi-competitive learning mechanism of Competitive and Cooperative Learning (CCL), this paper presents a new robust learning algorithm named Cooperation Controlled Competitive Learning (CCCL), in which the learning rate of each seed points within the same cooperative team can be adjusted adaptively. CCCL has not only inherited the merits of CCL, RPCL and its variants, but also overcome most of their shortcomings. It is insensitive to the initialization of the seed points and applicable to the heterogeneous clusters with an attractive accurate convergence property. Experiments have shown the efficacy of CCCL. Moreover, in some case its performance is prior to CCL and some other variants of RPCL.