摘要
在网络入侵异常检测中,数据预处理是一个非常重要的步骤,数据预处理的好坏直接影响后续检测的准确性.本文针对基于层次聚类的网络入侵异常检测中两个问题,在数据预处理阶段做出改进,一是属性冗余和属性权重问题,运用粗集理论对各个属性赋予权重并进行属性约减,二是粗集理论中连续数据离散化问题,提出了针对数据特点的自适应离散化算法,该算法是根据样本属性值分布来决定离散间隔,最后针对两个改进方法进行了实验,并与采用现有离散化方法进行了对比,实验结果证明了该算法的有效性和准确性.
In the network intrusion anomaly detection, data preprocessing is a very important component, it impacts on the accuracy of the follow - up detection. There are two problems in the network intrusion anomaly detection based on hierarchical clustering to solve in the preproeessing stage. First, It is attribute redundant and attribute weight problem. Using rough set theory computes the each attributes weights and reduces the attributes; Second, It is data discretion in rough set theory. This paper presents an adaptive discretization algorithm for data features. The algorithm determines the discrete intervals by the distribution of the value of the object＇s properties. Finally, experiments are conducted for these two improved methods, and compared with other representation discretization algorithms. The experimental results prove the validity and accuracy of the algorithm.
出处
《天津理工大学学报》
2013年第6期31-35,共5页
Journal of Tianjin University of Technology
基金
国家自然科学基金(61170174)
关键词
网络入侵异常检测
聚类
数据预处理
离散化
network intrusion anomaly detection
clustering
data preprocessing
diseretization
