基于大数据的K-means聚类算法在网络安全检测中的应用  

The Application of K-means Clustering Algorithm Based on Big Data in Network Security Detection

作　　者：郑志娴[1] 王敏[1] 

机构地区：[1]福建船政交通职业学院信息工程系,福州350007

出　　处：《湖北第二师范学院学报》2016年第2期36-40,共5页Journal of Hubei University of Education

基　　金：福建省教育厅B类科技项目(JB13292)

摘　　要：随着云时代的到来,大数据的应用受到了越来越多的关注。大数据的核心在于挖掘数据中蕴藏的价值链,为决策提供可借鉴参考。聚类算法是数据挖掘的一种归类方法,K-means则是基于划分的聚类方法。在网络安全检测中,应用K-means建立网络异常检测模型,可有效提高大数据环境下集中选取数据准确性的能力,控制检测误报率,缩短网络异常数据选取时间。但是,传统的K-means聚类算法在数据类型预处理、初始中心选取和K值确定等方面存在不确定性,导致对入侵检测的效率降低。本文提出一种改进的K-means算法,并通过利用KDDCup99数据集进行仿真实验,证明改进后的K-Means算法的检测准确率与检测效率要优于传统算法。With the advent of the cloud era,the application of big data has attracted increasingly great attention. The core of big data is to mine the hidden value chain in data,thus providing reference for decision-making. The clustering algorithm is a classification method of data mining,K-means is a clustering algorithm based on partition. In the network security detection,the establishment of network anomaly detection model based on K-means can effectively improve the ability to centrally select data with accuracy in large data environment,control the detection false alarm rate and shorten the time for network anomaly data selection.However,the traditional K-means clustering algorithm has its uncertainty in the aspects of the selection of data type pretreatment,the selection of initial center and determination of K value and so on,which leads to lower efficiency in intrusion detection. This paper proposes an improved K-means algorithm, and uses the KDDCup99 data set to carry out the simulation experiment,which proves that the improved K-Means algorithm is better in the detection accuracy and detection efficiency than the traditional algorithm.

关 键 词：大数据 网络安全检测 K-MEANS聚类算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]