摘要
文中首先设计出一类双曲函数型神经网络 ,提出一种层次双曲函数型网络 ,分析了其逼近特性 ,给出了其整体学习算法 ,使之更适合于具有层次知识结构背景的应用领域。因此 。
In this paper,a kind of function link artificial neural networks are designed and a new hierarchical function link artificial neural network is presented.Its universal approximation property is discussed,and the universal learning algorithm is presented.The network is suitable for application areas with hierarchical knowledge structures,and therefore has broad application prospects.[WT5HZ]
出处
《计算机工程与科学》
CSCD
2001年第3期59-61,69,共4页
Computer Engineering & Science
