基于无监督学习的入侵流量检测分类

方君1; 王茜2; 孙雪丽2

基于无监督学习的入侵流量检测分类

引用本文：方君1，王茜2，孙雪丽2.基于无监督学习的入侵流量检测分类[J].计算技术与自动化,2025,(2):1-8

摘要点击次数: 1220

全文下载次数: 848

作者	单位
方君1，王茜2，孙雪丽2	（1. 海军航空大学信息融合研究所,山东烟台 264001；2. 海军航空大学航空基础学院,山东烟台 264001）

中文摘要:针对现有入侵流量检测模型对小样本分类准确率低的问题，提出了一种基于Wasserstein Divergence Objective for GANs (WGAN div)和Information Maximizing Generative Adversarial Nets(Info GAN)的无监督学习入侵流量分类模型。首先，通过对不平衡的数据训练集进行过采样改善数据分布，然后对非数据部分进行独热编码处理并与数据部分整合，降低预处理复杂度，最后利用Info GAN模型进行数据训练，并在NSL KDD、CICIDS2017、UNSW NB15数据集进行性能评估和算法效能对比。实验结果表明，算法在多分类任务准确率分别达到91.0%、97.1%、79.9%，二分类任务准确率可达90.9%、96.9%、86.1%。相比于经典深度学习算法，Info GAN模型的准确率更高，误报率更低，具备较高的可靠性和工程应用价值。

中文关键词:入侵流量检测生成对抗网络过采样不平衡数据集

Intrusion Traffic Detection and Classification Based on Unsupervised Learning

Abstract:To solve the problem that the classification accuracy of model small samples is low, an unsupervised learning intrusion traffic classification model based on Wasserstein divergence objective for GANs (WGAN div) and Information Maximizing Generative Adversarial Nets(Info GAN)is presented. Firstly, the unbalanced data training set is oversamped to improve the data distribution. Then, the non data part is processed by independent thermal coding and integrated with the data part to reduce the complexity of pretreatment. Finally, the Info GAN model is used for data training.Performance evaluation and algorithm efficiency comparison were carried out in NSL KDD, CICIDS2017 and UNSW NB15 data sets.The experimental results show that the accuracy of multi classification task is 91.0%, 97.1%, 79.9% respectively, and the accuracy of binary classification task is 90.9%, 96.9%, 86.1% respectively.Compared with the classical deep learning algorithm, the Info GAN model has higher accuracy and lower false positive rate, and has higher reliability and engineering application value.

keywords:intrusion traffic detection generative adversarial nets oversampling unbalanced datasets

查看全文 查看/发表评论 下载pdf阅读器