改进的进化神经网络算法及其在入侵检测中的应用

Improved Evolutionary Neural Network Algorithm and Its Applications in Intrusion Detection

改进的进化神经网络算法是采用双种群的进化规则,同时完成对权值和结构的进化,其特点是加快算法的收敛速度,在一定程度上克服了BP算法陷入局部最小点的不足。将该算法应用于入侵检测领域中,建立一个基于改进的进化神经网络入侵检测系统模型,并用KDDCUP99数据测试了该模型中改进的进化神经网络分类器引擎,与基于BP神经网络和传统的进化神经网络等相比,得到了较高的检测率。

The improved evolutionary neural network algorithm can evolve neural network architectures and weights simultaneously using the evolutional rule of bi - population. It can resolve the shortage of existing least part point of BP neural network. The performance and convergent precision of the improved evolutionary neural network algorithm is improved greatly. A study of application of the algorithm in intrusion detection is proposed,and establish an intrusion detected system model based on improved evolutionary neural network is established. Using KDDCUP99 data set to test assorting machine model of evolutional neural network in this model. Experiment result shows that the method gains more higher detected ratio than the method based on BP neural network and the traditional evolutionary neural network.