神经网络结构的动态构筑方法  

A Dynamic Construction Method of Neural Network

作　　者：张宇[1] 熊善清[1] 王海燕[1] 

机构地区：[1]湖北工学院,湖北武汉430068

出　　处：《微机发展》2004年第1期32-33,共2页Microcomputer Development

摘　　要：如何决定人工神经网络的适当规模,以往都是通过试探法实现,不但费时,而且无规律可循。文中基于神经网络的基本学习算法,构筑动态网络结构,使之更符合抽取的新的输入、输出特性。讨论了构筑动态神经网络的一种途径。学习是从最简单的网络(无隐含单元)开始,新的单元一步一步补充,直到网络给出一个满意的模拟值。How to determine the appropriate size of an artificial neural network, it has usually been done through educated guess, which either wastes time or has no rule. In this paper , based on learning algorithm of neural network, dynamic neural network corresponding to extracting new features from the input and output attribute is constructed.A way for building dynamic neural network is discussed.It starts from the most simple network for the learning.The new units are added step by step,and get a satisfied simulation value finally.

关 键 词：人工神经网络 学习算法 动态构筑方法 网络结构 隐含层单元 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]