摘要
本文对BP神经网络算法进行了讨论,并通过实例分析了几种改进算法的实用性.结果表明,适当的改进动量因子和隐单元个数将加快网络的收敛性,提高预测能力.
We give some discuss to BP arithmetic of neural network, and give analyses to practicability of several improved arithmetic by some instances. The result presents that improve properly momentum gene and the number of concealed cells will quicken the astringency of network and increase ability of forecast.
出处
《中央民族大学学报（自然科学版）》
2008年第4期35-40,共6页
Journal of Minzu University of China(Natural Sciences Edition)
关键词
人工神经网络
BP算法
动量因子
artificial neural network
BP algorithms
momentum gene
