6 The Radial Basis Function Network

Compared to the feedforward network, the radial basis function (RBF) network is the next-most-used network model. As the name implies, this network makes use of radial functions. Section 2.5.2, Radial Basis Function Networks gives a tutorial introduction to this network model.

Feedforward networks and RBF networks can be used for the same types of problems, and the commands and their options are very similar. Therefore, instead of repeating a lot of information, the presentation here will be brief, giving references to the corresponding place in the section on feedforward networks where possible.

Using the symbolic computing capability of Mathematica, the structure of the RBF network may be modified much more naturally than is possible in other neural network software tools. Some of the features are listed in Chapter 5, The Feedforward Neural Network.