A NEW MODEL ON RADIAL BASIS PROBABILISTIC NEURAL NETWORKS （Ⅱ）:CASE STUDIES

The mapping properties of the radial basis probabilistic neural network model （RBPNN） are analyzed in detail from the viewpoint of linear algebra. The one dimensional cross images of five types of aircrafts obtained in a microwave darkroom are used to verify the classification performances of the proposed network. The experimental results show that this new model is very effective and feasible for labeled pattern classification.