基于Stacking模型融合的网络异常事件监测  

Network Abnormal Event Monitoring Based on Stacking Model Fusion

作　　者：张建东[1,2] 刘才铭[1,2] 李勤[1,2] ZHANG Jiandong;LIU Caiming;LI Qin(School of Electronic Information and Artificial Intelligence,Leshan Normal University,Leshan Sichuan 614000,China;Intelligent Network Security Detection and Eval uation Laboratory,Leshan Normal University,Leshan Sichuan 614000,China)

机构地区：[1]乐山师范学院电子信息与人工智能学院,四川乐山614000 [2]乐山师范学院网络安全智能检测与评估实验室,四川乐山614000

出　　处：《乐山师范学院学报》2022年第8期44-47,共4页Journal of Leshan Normal University

基　　金：互联网自然语言智能处理四川省高等学校重点实验室项目(INLP201909)。

摘　　要：随着云计算和人工智能的飞速发展,各种智能化的服务使大量数据在网络节点之间传输。然而,网络威胁,网络攻击和网络入侵在各种网络环境中大量增加。基于数据科学和机器学习,文章使用一系列涉及数据预处理、新特征创建、集成学习、Stacking模型融合等技术,以识别可疑的网络事件。提出了一种Stacking模型融合方法以提高网络异常事件识别的准确率,其中单一的模型包括XGBoost、Cat Boost、Light GBM、Random Forest。通过整合机器学习和数据科学相关技术,以提高网络异常事件识别率,实验结果表明该方法比单一的机器学习模型有更高的准确率。With the rapid development of cloud computing and artificial intelligence,Large amount of data of various intelligent services is transmitted between network nodes.However,cyber threats,cyber attacks and cyber intrusions have increased rapidly in various cyber environments.Based on data science and machine learning,this paper uses a series of techniques involving data pre-processing,new feature creation,ensemble learning,stacking model fusion,etc.to identify suspicious network events.A stacking model fusion method is proposed to improve the accuracy of network abnormal event recognition,and the single model includes XGBoost,CatBoost,LightGBM,RandomForest.By integrating machine learning and data science related technologies to improve the identification rate of abnormal network events,the experimental results show that this method has a higher accuracy than single machine learning model.

关 键 词：网络异常事件 集成学习 网络安全 机器学习 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]