计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (3): 68-75.DOI: 10.3778/j.issn.1002-8331.1711-0100

• 网络、通信与安全 • 上一篇    下一篇

基于概念漂移检测的自适应流量分类方法

姜振东1,王建明1,潘吴斌2   

  1. 1.南京工业大学 计算机科学与技术学院,南京 211816 
    2.东南大学 计算机科学与工程学院,南京 210096
  • 出版日期:2019-02-01 发布日期:2019-01-24

Adaptive Traffic Classification Approach Based on Concept Drift Detection

JIANG Zhendong1, WANG Jianming1, PAN Wubin2   

  1. 1.School of Computer Science and Technology, Nanjing University of Technology, Nanjing 211816, China
    2.School of Computer Science and Engineering, Southeast University, Nanjing 210096, China
  • Online:2019-02-01 Published:2019-01-24

摘要: 针对网络流特征会随网络环境变化而发生改变,从而导致基于流特征的机器学习分类方法精度明显降低的问题。提出一种基于概念漂移检测的自适应流量分类方法,该方法借助Kolmogorov-Smirnov检验对出现的流量进行概念漂移检测,然后通过多视图协同学习策略引入新流量样本修正概念漂移导致的模型变化,使分类器得到有效更新。实验结果表明该方法可以有效检测概念漂移并更新分类器,表现出较好的分类性能和泛化能力。

关键词: 概念漂移, Kolmogorov-Smirnov检验, 协同学习, 流量分类

Abstract: For network traffic characteristics will change with the change of network environment, which causes a significant reduction on the accuracy of traffic classification method based on machine learning. This paper proposes an adaptive traffic classification method based on concept drift detection, the method adopts Kolmogorov-Smirnov test to detect traffic concept drift, and then the classifier to be effectively updated through multi-view cooperative schema by introduction of new coming traffic in response to revise the model change caused by concept drift. Experimental results show that the method can effectively detect concept drift and update the classifier it has high accuracy and generalization.

Key words: concept drift, Kolmogorov-Smirnov test, cooperative learning, traffic classification