Abstract
Restricted Coulomb Energy(RCE) network is one of the competitive learning networks that are able to classify input data, together with the Self-Organizing Map (SOM) and the Learning Vector Quantization(LVQ). In the RCE network, there is no need for setting the number of required neurons before learning because the RCE network makes new neurons automatically to classify input data into correct categories. In this report, we propose a new RCE model and its network with RBF output function in order to reduce the number of neurons created in the network, and evaluate the classification efficiencies by solving the iris problem and the shuttle problem. Moreover, we investigate the abilities of the RCE model concerning the incremental learning unsolvable in LVQ. We find that our proposed model can perform in classification accuracies with smaller number neurons to compete with the conventional ones.
