摘要
针对现代大数据环境中网络入侵检测系统(network intrusion detection system,NIDS)难以应对复杂网络攻击的问题,提出了一种基于深度神经网络(Deep Neural Network,DNN)-门控循环单元(Gated Recurrent Unit,GRU)-支持向量机(Support Vector Machine,SVM)的组合模型DNN-GRU-SVM。该模型结合了DNN、GRU与SVM的优势,首先利用DNN提取网络数据特征,通过调整学习率与批量归一化来加速训练并减少过拟合;采用GRU捕捉序列数据中的时间依赖性;通过SVM实现精确分类。在KDD Cup'99数据集上的实验表明,DNNGRU-SVM组合模型取得了显著的性能提升,其检测准确率达94.53%,精确度为99.8%,召回率为92.8%,F1分数为96.2%,显著优于传统机器学习算法及单一的深度神经网络。实验结果表明,该模型能够有效提高网络入侵检测的准确率和适应性,为复杂网络环境下的入侵检测提供了可靠的解决方案。
A combined model named DNN-GRU-SVM,based on Deep Neural Network(DNN),Gated Recurrent Unit(GRU),and Support Vector Machine(SVM),was proposed to address the issue that network intrusion detection systems(NIDS)struggled to cope with complex network attacks in modern big data environments.This model integrated the advantages of DNN,GRU,and SVM.Firstly,DNN was utilized to extract features from network data,accelerating training and reducing overfitting by adjusting the learning rate and applying batch normalization.GRU was employed to capture the temporal dependencies in sequence data.SVM was then used to achieve precise classification.Experiments conducted on the KDD Cup'99 dataset demonstrated that the DNN-GRU-SVM combined model achieved significant performance improvements,with a detection accuracy rate of 94.53%,precision of 99.8%,recall of 92.8%,and F1 score of 96.2%,significantly outperforming traditional machine learning algorithms and single deep neural networks.The experimental results indicated that this model could effectively enhance the accuracy and adaptability of network intrusion detection,providing a reliable solution for intrusion detection in complex network environments.
作者
刘虎鹏
颜辉
于萍
许晓晴
龙蕴鑫
耿晓中
龙多
赵禺
LIU Hupeng;YAN Hui;YU Ping;XU Xiaoqing;LONG Yunxin;GENG Xiaozhong;LONG Duo;ZHAO Yu(Jilin Province S&T Innovation Center for Physical Simulation and Security of Water Resources and Electric Power Engineering,Changchun Institute of Technology,Changchun 130012,China;Information Engineering College,Suqian University,Suqian 223800,China;Changchun Xinheng Optoelectronic Information Technology Co.,Ltd.,Changchun 130000,China;Changchun University of Chinese Medicine,Changchun 130117,China)
出处
《电脑与信息技术》
2025年第4期64-70,共7页
Computer and Information Technology
基金
吉林省科学技术厅项目(No.20240404067ZP)。
