基于VB的RBF神经网络编程及验证

Programming and verification of RBF neural network with VB

简要介绍了RBF(径向基函数)神经网络的基本原理和编程过程,通过函数逼近的例子以及与BP(误差反向传播)神经网络的函数逼近比较分析,说明RBF神经网络的泛化能力要优于BP神经网络,能够更快、更准确地逼近函数;同时也说明RBF神经网络相对BP神经网络需要更复杂的网络结构来保证相同精度。试验结果证实了用VB开发的RBF神经网络软件正确、可用。

In this paper,the principle of RBF neural network and its programming are briefly introduced.Through the examples of function approximation and comparison with BP neural network,it is shown that RBF neural networks have better generalization ability than BP neural networks and can approach functions faster and more accurately.However,compared to BP neural networks,RBF neural networks require a more complex network structure to ensure the same accuracy.The experimental results confirm that the RBF neural network software developed by VB is correct and usable.