标题:On the combination of simulated annealing and semi-supervised clustering for intrusion detection
作者:Feng, Guorui ;Wu, Jian
通讯作者:Feng, Guorui
作者机构:[Feng, Guorui ;Wu, Jian ] Evidence Forensic Laboratory in Colleges and Universities of Shandong Province, Jinan, China;[Feng, Guorui ;Wu, Jian ] Schoo 更多
会议名称:4th International Conference on Electronics, Communications and Networks, CECNet2014
会议日期:12 December 2014 through 15 December 2014
来源:Electronics, Communications and Networks IV - Proceedings of the 4th International Conference on Electronics, Communications and Networks, CECNet2014
出版年:2015
卷:2
页码:1389-1393
关键词:Intrusion detection; Semi-supervised K-means clustering; Simulated annealing
摘要:The traditional intrusion detection system based on clustering has the problems of high misdetection rate and low detection rate with the lack of supervised data. According to this situation, this paper proposes an intrusion detection method combining the simulated annealing algorithm with the semi-supervised K-means. First, it improves the cluster initialization by a small amount of labeled data for network intrusion; then the idea of semi-supervised learning is introduced to build the K-means algorithm. The simulated annealing algorithm has the ability of jumping out of the local optimal solution. The combination of the semi-supervised K-means clustering and the SA algorithm can realize global optima. The method shows the clusters with labeled data, which can be used for intrusion detection. Based on the KDDCUP99 data set, the experiment implies that the method can highly improve the clustering algorithm performance and the precision rate of intrusion detection than the traditional k-means algorithm or the semi-supervised K-means algorithm. © 2015 Taylor & Francis Group, London.
收录类别:EI;SCOPUS
资源类型:会议论文;期刊论文
原文链接:https://www.scopus.com/inward/record.uri?eid=2-s2.0-84960193588&partnerID=40&md5=a2d08ae208e5a306d3e28ab2e5d35a7d
TOP