基于Boosting的网络异常流量检测算法研究  

A Study of Network Abnormal Traffic Detection Algorithm Based on Boosting

作　　者：于振洋[1] 

机构地区：[1]淮阴工学院计算机工程学院,江苏淮安223003

出　　处：《淮阴工学院学报》2011年第5期39-43,共5页Journal of Huaiyin Institute of Technology

摘　　要：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。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 weak learning algorithm,Boosting is effective in non-supervision cases.

关 键 词：机器学习 网络流量检测 BOOSTING 支持向量机 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]