贝叶斯分类器在入侵检测中的应用

本文针对贝叶斯网络分类器的时间复杂性均取决于特征值间的依赖程度,甚至可能是NP完全问题,提出了一种基于分治理论的贝叶斯分类器算法,该算法把分治理论的思想和贝叶斯网络分类器有机结合起来,既保留了贝叶斯网络分类器模型的结构简单、复杂度低的优点,又降低了传统贝叶斯分类器时间复杂度的问题。通过大量数据的进行贝叶斯网络训练得到的检测结果,与传统的基于神经网络和遗传算法相比,检测率提高5%以上。同时,该算法还有误检率低的优点。

The Application Of Bayesian classifier in intrusion de-tection

To solve the problem that time complexity of the Bayesian classifier was determined by the degree of dependence of characteristic values which was maybe a NP completed problem,we proposed an Bayesian classifier algorithm based on the partitioning theory the Bayesian classifier algorithm not only reserved the advantages of simple structure and low complexity of traditional.Bayesian classifier but also highly lowered the time complexity of traditional Bayesian classifier.Compared with the algorithms based on ...