Optimal learning for patterns classification in RBF networks

The proposed modifying of the structure of the radial basis function (RBF) network by introducing the weight matrix to the input layer (in contrast to the direct connection of the input to the hidden layer of a conventional RBF) so that the training space in the RBF network is adaptively separated by the resultant decision boundaries and class regions is reported. The training of this weight matrix is carried out as for a single-layer perceptron together with the clustering process. This way the network is capable of dealing with complicated problems, which have a high degree of interference in the training data, and achieves a higher classification rate over the current classifiers using RBF.