A Concurrent Fuzzy Neural Network Approach for a Fuzzy Gaussian Neural Network

The aim of the paper is to introduce a concurrent fuzzy neural network approach, representing a winner-takes-all collection of fuzzy Gaussian modules. Our proposed model will be applied for the pattern classification. The fuzzy neural model consists of a set of M fuzzy neural networks, one for every class, each network having a single output. The output value corresponding to the M k k , 1 ,  neural network is equal to 1 for those patterns belonging to the class k and 0 for the others patterns from the training set. After we have trained the M fuzzy neural networks, we shall save the weights in M files, in order to be used in the test stage of the respective networks. We have applied this model as a classifier, in a cascade having the following processing stages: the application of a pattern to the input of each of the M networks; computing the output of the respective network and then taking the maximum of those M outputs. The results of computer simulation will be given.