RBF神经网络的函数逼近能力及其算法

FUNCTION APPROXIMATION CAPABILITY AND ALGORITHMS OF RBF NEURAL NETWORKS

BP网络是广为应用的一种前馈网络,但是由于其自身缺陷,其它网络越来越受到关注.目前RBF网络是前馈网络研究中的一个热点,有关其逼近理论以及算法的研究出现在许多文献之中.本文综述了RBF网络理论,对其结构、函数逼近性质、学习算法做了较为详细的介绍,并指出了RBF网络各种学习算法的优点和存在的问题.

BP network is a kind of feedforward network that is widely used. But because of its own defects, other networks are drawing more and more attentions. At present, RBF network becomes one of the focuses of research on feedforward networks. The research results on its approximation theory and algorithms have presented in a lot of literatures. In this paper, RBF network theory is surveyed with comparatively detailed introduction to its structure, property of function approximation, algorithms of studying. At the same time, the merits and the problems of various kinds of studying algorithms are pointed out.