The best of both worlds: Casasent networks integrate multilayer perceptrons and radial basis functions

Although multilayer perceptrons (MLPs) and radial basis functions (RBFs) appear to be quite different approaches to function approximation, a simple but profound insight by D. Casasent and E. Barnard (1990) has made it possible to completely unify two approaches. The authors complete the unification and comment on the potentially significant increase in representational power this Casasent network offers. They eliminate the distinction between MLP networks and RBF networks by unifying them into a single Casasent network that possesses all of their separate capabilities. New questions regarding learning methodologies for the Casasent network are also presented.<>