摘要
叙述了ＢＰ神经网络的基本思想，分析了ＢＰ网络产生振荡和平台的原因，给出了解决问题的方法．同时介绍了一个ＢＰ神经网络的数字识别系统（ＤＧＲＳ）．
The traiditional back propagation algorithm often produces oscillation and platform problems and affects the speed of convergence of the algorithm.The paper describes the basic thought of BP neural networks,analyses the cause of oscillation and platform problem produced by networks and puts forward methods of solution.It also introduces a digital recognition system of BP neural networks, the DGRS.
出处
《北京理工大学学报》
EI
CAS
CSCD
1995年第5期72-77,共6页
Transactions of Beijing Institute of Technology
关键词
神经网络
振荡/BP算法
平台
neural networks
oscillations/back propagation algorithm
platform
