基于SVM决策树的网络流量分类

提出一种用支持向量机(SVM)决策树来对网络流量进行分类的方法,利用SVM决策树在多类分类方面的优势,解决SVM在流量分类中存在的无法识别区域和训练时间较长的问题。对权威流量数据集进行了测试,实验结果表明,SVM决策树在流量分类中比普通的"一对一"和"一对多"SVM方法具有更短的训练时间和更好的分类性能,分类准确率可以达到98.8%。

Network Traffic Classification Using SVM Decision Tree

In order to solve the unrecognized area and long training time problems existed when using Support Vector Machine (SVM) method in network traffic classificationSVM decision tree was used in network traffic classification by using its advantages in multi class classification.The authoritative flow data sets were tested.The experiment results show that SVM decision tree method has shorter training time and better classification performance than ordinary “one on one” and “one on more”SVM method in network traffic classificationwhose classification accuracy rate can reach 98.8%.