论文部分内容阅读
网络高度发展的今天,网络安全已经提升到了一个空前的高度,入侵检测系统应用的兴起也促进了更多的入侵检测技术与算法的研究。检测系统的监测性能直接取决于对训练样本集中数据样本的学习。而现阶段构建训练样本集又十分的昂贵和费时。支持向量机SVM是一种能在训练样本数很小的情况下达到很好的分类推广学习算法,它能够较好地解决小样本学习问题,同时具有很好的泛化能力。本文提出将支持向量机主动学习运用于入侵检测,通过SVM主动挑选学习样本,在保证分类器的分类精度不降低的情况下,达到提高训练速度和降低构建训练样本集成本的目的。
Today, with the development of network, network security has been raised to an unprecedented height. The rise of application of intrusion detection system has also promoted more intrusion detection technologies and algorithms. The monitoring performance of the detection system directly depends on the learning of the data samples concentrated on the training samples. At this stage to build training sample set and very expensive and time-consuming. Support Vector Machine (SVM) is a classified generalized learning algorithm that can achieve good training samples in a very small number of training samples. It can solve the small sample learning problem well and has good generalization ability. In this paper, the active learning of SVM is applied to intrusion detection, and the learning samples are selected by SVM initiatively. In order to improve the training speed and reduce the cost of constructing the training samples, the classification accuracy of the classifier is not reduced.