A new neural network model based on nearest neighbor classifier

A new neural network model for pattern classification based on the nearest neighbor method is presented. In this model, the training patterns were mapped to hidden neurons, but one hidden neuron may represent one or more training patterns. At the recognition stage, the distance to the training patterns were calculated in parallel by the hidden neurons. Therefore, the nearest neighbor can be found efficiently. Some experimental results on recognition of printed and hand-written numerals are given to evaluate the proposed model. Comparisons with the backpropagation learning algorithm are included