ABSTRACT
This chapter considers radial basis function (RBF) networks. A RBF network can be described as a parametrized model used to approximate an arbitrary function by means of a linear combination of basic functions. RBF networks belong to the class of kernel function networks where the inputs to the model are passed through kernel functions which limit the response of the network to a local region in the input space for each kernel or basis function. The chapter describes learning algorithms for RBF networks and shows how they can overcome the problem of training that normally occurs in multilayer networks. It explores the use of RBF networks in terms of basic theory, architectures, learning algorithms, and applications to signal processing. An interesting relationship that can be drawn between RBF networks and probabilistic framework is that when a Gaussian basis function is used, the model can be viewed as a mixture of normal density functions.
