基于神经网络的入侵检测方法研究

在传统BP算法的基础上,提出对BP算法的改进方法,并对KDDCup99数据集进行预处理,以达到提高检测效率、降低误检率的效果。利用决策树ID3算法在筛选属性方面的优势,对KDDCup99数据集属性进行筛选,并将结果进行标准化与归一化处理,使训练数据和测试数据更加符合网络训练的数值特征。最后通过Matlab进行编程,将结果与传统BP算法加以比较,证实了该方法的有效性。

Research on the Methods of Intrusion Detection Based on Neural Network

According to the analysis of the BP algorithm, this article proposes the impoved method to BP algorithm. Meanwhile, by preprocessing the KDDCup99 Data Set, it could increase detection rate and detection efficiency and reduce false detection rate. Due to the decision tree ID3 algorithm＇s advantage in attribute filtration, we use this algorithm in the KDDCup99 data set to filter the attributes. Then, the results got standarded and normalized, in the order to adapt training data and testing data to the numerical characteristics. In the final step, we simulated by software Matlab. Through the comparation of the experimental result and the one of traditional BP algorithm, the proposed method was proved approprivate.