信息网络安全 ›› 2018, Vol. 18 ›› Issue (1): 9-14.doi: 10.3969/j.issn.1671-1122.2018.01.002

• • 上一篇    下一篇

AR-HELM算法在网络流量分类中的应用研究

魏书宁1,2(), 陈幸如1,2, 唐勇3, 刘慧1,2   

  1. 1.湖南师范大学物理与信息科学学院,湖南长沙410006
    2.湖南师范大学物联网技术及应用重点实验室,湖南长沙410006
    3.国防科技大学计算机学院,湖南长沙410073
  • 收稿日期:2017-12-05 出版日期:2018-01-20 发布日期:2020-05-11
  • 作者简介:

    作者简介:魏书宁(1979—), 女,湖南,副教授, 博士,主要研究方向为智能控制、数据分析等;陈幸如(1994—),女,安徽,硕士研究生,主要研究方向为网络流量分类、大数据分析;唐勇(1979—),男,湖南,副研究员,博士,主要研究方向为网络安全;刘慧(1982—),女,湖南,实验师,硕士,主要研究方向为数字图像处理。

  • 基金资助:
    国家自然科学基金[61472437];湖南师大自然科学研究项目[物160432]

Research on the Application of AR-HELM Algorithm in Network Traffic Classifi cation

Shuning WEI1,2(), Xingru CHEN1,2, Yong TANG3, Hui LIU1,2   

  1. 1. College of Physics and Information Science, Hunan Normal University, Changsha Hunan 410006, China
    2. Internet of Things Technology and Application Key Lab, Hunan Normal University, Changsha Hunan 410006, China
    3. College of Computer, National University of Defense Technology, Changsha Hunan 410073, China
  • Received:2017-12-05 Online:2018-01-20 Published:2020-05-11

摘要:

针对传统分类算法建模速度慢、精确度低、分类效率不理想等问题,一种基于粗糙集属性约简的极限学习机网络流量分类方法成为利用机器学习研究网络流量分类的热门方法。由于结构限制,一些特殊的自然信号数据使用极限学习机进行特征学习一定程度上并不是很有效。因此,文章提出一种基于改进的粗糙集属性约简的多层极限学习机算法作为分类算法进行建模。实验结果显示,相较传统的神经网络和机器学习算法,文章算法可以很好地应用于网络流量分类且改善了极限学习机的学习表现。改进后的算法模型获得了更快、更优质的聚合结果。

关键词: 网络流量分类, 属性约简, 极限学习机, 粗糙集

Abstract:

Considering the huge time overheads, low accuracy rate and undesirable classification efficiency of the conventional classification algorithms, extreme learning machine network traffic classification methods based on attribute reduction in rough set become hot methods which study network traffic classification using machine learning. Due to structural constraints, feature learning using extreme learning machine (ELM) may be ineffective for some special natural signal data. Thus, an improved hierarchical extreme learning machine algorithm based on attribute reduction in rough set (AR-HELM) is proposed as classification algorithm to construct model. The experimental results show that, comparison with traditional neural network and machine learning algorithm, the AR-HELM can be well applied to network traffic classification and improve the learning performance of the extreme learning machine. The improved algorithm model gets faster and better convergence results.

Key words: network traffic classification, attribute reduction, ELM, rough set

中图分类号: