迭代优化SVM网络流量分类技术研究  

Network traffic classification by iterative tuning support vector machine

作　　者：林海涛[1] 陈源[1] LIN Hai-tao CHEN Yuan(College of Electronic Engineering, Naval Univ. of Engineering, Wuhan 430033, Chin)

机构地区：[1]海军工程大学电子工程学院,武汉430033

出　　处：《海军工程大学学报》2016年第5期10-14,19,共6页Journal of Naval University of Engineering

基　　金：国家863计划资助项目(2009AA01Z205);国家重点科学专项基金资助项目(2010ZX03003-001)

摘　　要：针对传统SVM分类方法训练速度较慢的特点,提出了一种新的迭代调谐方法。该方法通过对SVM分类方法的参数进行迭代化以提高它的训练速度。从NetFlow数据中提取流层面信息进行流量分类的实验结果表明:迭代优化SVM分类的训练速度比传统8种SVM分类方法更快,同时保持和其他8种分类方法近乎相同的分类精度。A novel iterative tuning scheme is proposed to increase the training speed of the classification algorithm by support vector machine（SVM）.The equations to obtain SVM parameters are derived by theoretical analysis of iterative-tuning SVM.Traffic classification is carried out by using flow-level information extracted from NetFlow data.Performance evaluation demonstrates that the proposed iterative-tuning SVM exhibits a training speed that is two to ten times faster than those of eight other previously proposed SVM techniques,while it maintains almost the same classification accuracy as those SVM techniques.

关 键 词：流量分类 机器学习 SVM 训练速度 迭代优化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]