摘要
针对入侵检测数据高维且不均衡的问题,提出基于欠采样和对抗自编码器的入侵检测算法。首先,采用改进的EasyEnsemble欠采样方法将多数类样本多次采样分成多个子样本,训练多个子分类器,最终得到强分类器来处理数据不均衡问题,然后利用对抗自编码器对处理后的数据进行降维,最后用随机森林算法对处理后的新数据进行分类,来检测出高维且不平衡数据中的恶意攻击。实验结果表明,该算法相对于传统算法表现出较优的性能,能够有效地提高入侵检测的准确性,降低误报率。
Aiming at the problem of high dimensional and unbalanced intrusion detection data,an intrusion detection al gorithm based on adversarial auto-encoder was proposed.Firstly,undersampling method is used to deal with the problem of unbalanced data,then the adversarial autoencoder is used to reduce the dimension of the processed data,and finally the new data is classified by the random forest algorithm.Experimental results show that the algorithm can effectively improve the accuracy of intrusion detection and reduce the false alarm rate.
作者
郭文婷
张军
魏洪伟
刘莹
Guo Wenting;Zhang Jun;Wei Hongwei;Liu Ying(school of computer science and information engineering,Harbin normal university,Harbin 150025,China)
出处
《信息通信》
2019年第12期58-60,共3页
Information & Communications
基金
黑龙江省高等教育教学改革研究一般研究项目(S JGY20170180)
国家级大学生创新创业训练计划项目(201710231020).
关键词
特征降维
对抗自编码器
欠采样
不平衡数据
入侵检测
Feature dimension reduction
Adversarial autoencoder
Undersampling
Unbalanced data
Intrusion detection
