Considerations on two-layer radial based function network

This report presents the performance analysis of a two-layer radial based function network. The network has been developed as a new structure with new training methods. The basic distance measure is a standard Euclid distance. By using this distance, a conventional self-organization is applied to both layers. In addition, a backpropagation mechanism is also introduced into them in order to improve the accuracy of clustering and evaluation. Both training mechanisms are alternatively used to two layers. The developed network is compared with a conventional multilayer feedforward network and a self-organized network.