摘要
针对反向传播人工神经网络(BP ANN)收敛速度慢的问题,提出一种改进的自适应动量因子的BP算法。该算法使得权值向量和偏置值都可以根据自适应动量进行调整,减小了学习过程中出现振荡的趋势,缩短了网络收敛时间。基于改进BP ANN建立电子政务内网神经网络评估模型,比较改进前与改进后BP ANN评估模型的性能。仿真结果表明,改进后BP ANN评估模型收敛性能更稳定且评估误差更小,能够更好地为复杂的电子政务内网进行信息安全评估。
Aiming at the back propagation artificial neural network's problems of slow convergence rate,an improved back- propagation algorithm with adaptive momentum factor is proposed. With this algorithm,the training speed of the neural networks,is improved and the oscillation trend appeared in the process of learning is reduced. An evaluation model is proposed by using an artificial neural network based on improved BP model. The comparison between the origin and the improved are indicates that the improved BP ANN model's convergence performance is more stable and evaluation error is smaller.The experimental results demonstrate that the proposed BP ANN evaluation model can apply to evaluate E- government network's information security.
出处
《北京电子科技学院学报》
2014年第2期35-40,共6页
Journal of Beijing Electronic Science And Technology Institute
基金
"中央高校基本科研业务费资助(项目编号:2014CLJH05)"
关键词
BP神经网络
自适应动量因子
电子政务内网
信息安全
安全评估
Back propagation neural network
Adaptive momentum factor
Electronic Government
information security
security assessment
