基于Boosting的网络异常流量检测算法研究

从机器学习的角度系统研究网络流量检测,将Boosting算法引入到网络异常流量的检测当中,设计两种不同的弱学习方法：估计多变量高斯分布和估计超球体区域。实验结果表明,基于Boosting的检测算法性能要优于一类支持向量机,同时也表明作为一种提升弱学习算法性能的一般性策略,Boosting在非监督情况下是非常有效的。

A Study of Network Abnormal Traffic Detection Algorithm Based on Boosting

From the perspective of machine learning system,detecting network traffic is studied.Boosting algo-rithm is applied to the detection of network's abnormal traffic in the weak design of two different learning meth-ods: an estimated multivariate Gaussian distribution and the estimated super-sphere region.Experimental re-sults show that the detection algorithm based on Boosting is superior to the first class of support vector machines,which indicates that,as a general strategy to enhance the performance of wea...