摘要
为了克服传统误差反向传播算法收敛速度慢且容易陷入局部极小的问题,提出了一种改进的误差反向传播算法,并给出了一个基于神经网络的入侵检测系统的模型,阐述了该模型的设计思想.最后通过训练过程和检测过程对实验的结果进行了客观的分析.分析结果表明:改进的误差反向传播算法运用于神经网络入侵检测漏检率和误报率都比较高,而且对未知类型的攻击,也有一定的检测效果,说明改进的误差反向传播算法在神经网络入侵检测方面具有很大的发展空间和应用前景.
In order to overcome .the traditional error back - propagation algorithm with slow convergence and easy to fall into local minimum problem. This paper first introduces an improved error back - propagation algorithm, and give a neural network- based intrusion detection system model, expounded the idea of the model design. Finally, through the training process and testing process, an objective analysis was carried out on the results of experiments, which showed that： undetected intrusion detection rate and false alarm rate is higher, when the improved error back - propagation algorithm applied to neural network. It is also effective to types of attacks. It shows improved error back- propagation algorithm in neural network intrusion detection has ample space of the development and application prospects.
出处
《微电子学与计算机》
CSCD
北大核心
2009年第8期240-242,共3页
Microelectronics & Computer
关键词
改进的误差反向传播算法
神经网络
入侵检测
improved error back-propagation algorithm
neural network
intrusion detection
