摘要
本文评述了近十几年来国内外对前向神经网络设计问题的研究情况,在分析各种已有设计方法优缺点的基础上,提出另一种新的解决前向神经网络设计问题的方法,并给出几个非常典型的设计(模拟)例子,以说明本文所提出方法的有效性和潜力。
In this paper,the recent approaches to the design of feedforward neural networks are introduced,and their advantages and disadvantages are discussed as well.Based on the analysis,a new approach to the design problem is proposed.Several typical simulation examples are presented to show the potential usage of the new approach in practice.
出处
《计算机工程与科学》
CSCD
1998年第4期1-10,共10页
Computer Engineering & Science
基金
国家自然科学基金
关键词
前向神经网络
BP算法
径基函数网络
人工智能
feedforward neural networks,BP algorithm,radial basis function networks,covering algorithm.