Matlab神经网络十讲（6）：RBNN、GRNN与PNN

1. Introduction to Radial Basis Neural Networks

Radial basis networks typically need more neurons compared to standard feedforward backpropagation networks, yet they are often designed in a fraction of the time needed to train conventional feedforward architectures. These networks function optimally when large quantities of training data are accessible.

1.1 Important Radial Basis Functions

Neuron Model：

he transfer function for a radial basis neuron is
radbas(n) = exp(-n*n)

![](ht