6.2 Examples

This section gives some simple function approximation examples with one and two input variables and one and two output variables. The examples are analogous to the examples given on feedforward networks in Section 5.2. The reason for this similarity is that RBF networks and feedforward networks are used for the same type of problems. The best choice between these two alternative models depends on the specific problem. You can seldom tell which is best without trying them both.

Notice that if you reevaluate the examples, you will not receive exactly the same results. The reason for this indeterminacy is that the initial network parameters are partly random. See Section 6.1.1, InitializeRBFNet.