摘要: 为解决入侵检测中的数据约简问题,提出一种基于粗糙集的入侵检测数据约简算法,其中包括特征选择与属性值约简。特征选择部分采用互信息的方法消除冗余特征,属性值约简部分采用归纳值约简算法消除冗余属性值。实验结果表明,该方法不仅能缩短训练及检测时间,减小数据存储代价,还能提高分类精确度。
关键词:
粗糙集,
互信息,
入侵检测,
特征选择,
数据约简
Abstract: In order to solve the problems of data reduction in intrusion detection, this paper presents an algorithm of attack features selection based on mutual information including feature selection and value reduction. It uses mutual information method to eliminate the redundant features, and inductive value reduction algorithm to eliminate redundant attribute values. Experimental results demonstrate that it not only can reduce the cost of memory and time, but also can improve the accuracy of classification.
Key words:
rough set,
mutual information,
intrusion detection,
feature selection,
data reduction
中图分类号:
许晓东, 古一, 朱士瑞. 入侵检测中的数据约简研究[J]. 计算机工程, 2011, 37(11): 170-172.
HU Xiao-Dong, GU Yi, SHU Shi-Rui. Research on Data Reduction in Intrusion Detection[J]. Computer Engineering, 2011, 37(11): 170-172.