论文部分内容阅读
为了提高网络入侵检测系统(IDS)的实时性、可用性以及整体性能,提出了一种自动识别特征相关性的方法(特征分类法)。用该方法提取出的互相独立(或相关性很小)的特征作为反向传播神经元网络的输入,以此为基础建立了一个IDS。实验证明该方法以及所建立的IDS效果较好。结论表明通过分类可以求得一组特征互相之间的相关程度,进一步可求得互相独立(或相关性很小)的特征。
In order to improve the real-time, usability and overall performance of Network Intrusion Detection System (IDS), a method to automatically identify feature correlation (feature classification) is proposed. Using this method to extract independent (or less relevant) features as input to a backpropagation neural network, an IDS is built on this. Experiments show that this method and the established IDS better. The conclusion shows that the correlation between a group of features can be obtained by classification, and further independent (or less relevant) features can be obtained.