摘要
本文首先讨论了组合式前馈神经网络原理和一般设计方法, 然后用两个仿真实验展示了组合式前馈神经网络在处理函数逼近和模式分类问题上的有效性。
In this paper, we discuss the idea of linear combinations of neural networks. Experimental results based on simulated data are included. For our examples, the approxinmation accuracy and recognition effect resulting from using combining neural networks is better then single neural network.
出处
《长春光学精密机械学院学报》
1998年第3期39-43,共5页
Journal of Changchun Institute of Optics and Fine Mechanics
基金
国家兵器 "九五" 预研基金
关键词
前馈
神经网络
函数逼近
模式分类
组合式
Feedforward neural networks
Function approximation,
Pattern classification