基于机器学习的网络异常流量检测  

Network abnormal traffic detection based on machine learning

作　　者：沈徳松 SHEN Desong(Anhui Wenda University of Information Engineering,Hefei 231201,China)

机构地区：[1]安徽文达信息工程学院,安徽合肥231201

出　　处：《安徽科技学院学报》2024年第1期111-116,共6页Journal of Anhui Science and Technology University

基　　金：安徽省高校自然科学研究项目(2022AH052847)。

摘　　要：目的:探索基于XGBoost算法的网络异常流量检测方法,并评估其分类准确率。方法:根据X GBoost算法和主成分分析的技术原理,梳理了网络异常流量的类型、具体表现和异常流量的成因。采用136.4万条网络流量样本作为实验数据集,包括77个网络流量特征和8种网络流量类型。进一步构建XGBoost分类模型,采用多个分类器,实现对网络异常流量的有效检测和识别。结果:XGBoost算法对网络异常流量的检测准确率达到了96.32%。结论:XGBoost算法在网络异常流量检测方面具有出色的性能和可靠性,能够有效为网络管理员提供有效的辅助决策和保护措施。Objective:To explore the method of network abnormal traffic detection based on XGBoost algorithm and to evaluate its classification accuracy.Methods:Firstly,the XGBoost algorithm and the principle of principal component analysis were introduced,and the types,specific performance and causes of abnormal traffic were analyzed.Then,1364000 network traffic samples were used as the experimental data set,including 77 network traffic characteristics and 8 types of network traffic.Furthermore,the classification model of XGBoost was built to detect and identify the abnormal traffic effectively.Results:Experimental results showed that the detection accuracy of XGBoost algorithm for network abnormal traffic was 96.32%.Conclusion:The XGBoost algorithm had excellent performance and reliability in network abnormal traffic detection,and could provide effective assistant decision-making and protection measures for Network administrators.

关 键 词：机器学习 XGBoost算法 主成分分析 网络异常流量 

分 类 号：TP3-05[自动化与计算机技术—计算机科学与技术]