Practical implementation of a radial basis function network for handwritten digit recognition

A practical implementation of a radial basis function network for handwritten digit recognition is described. Inspired from regularization theory and Parzen windows nonparametric estimator, radial basis function networks are tested for a classification task. Reduction of the number of hidden nodes, which is an important and necessary step to obtain a computationally tractable network, is made using an original technique. A comparison is made with the k-nearest-neighbor method. Results appear better for the network at a much lower computational cost