基于熵重要测度权重粗糙集的阿尔法多层凝聚入侵分类

摘要

针对入侵检测数据量大，而文献[1]提出的α核心集的多层凝聚算法计算复杂度过高，影响实际应用的问题，提出一种基于熵重要测度权重粗糙集的α核心集多层凝聚入侵分类算法。首先，基于熵重要测度权重方法利用粗糙集对入侵检测数据进行预处理和属性约简，降低数据维数防止算法陷入“维数陷阱”；其次，用熵重要测度权重距离代替阿尔法多层凝聚算法的欧式距离计算个体相似度，并实现粗糙集输出数据与阿尔法多层凝聚算法的有效对接。通过实验表明，基于熵重要测度权重粗糙集的α核心集多层凝聚入侵分类算法能够更加有效对 KDD CUP 99标准数据库进行检测分类。%Intrusion detection has large data amount,and the multilevel aggregation clustering algorithm of αcore-set presented by literature [1]has too high computational complexity,which affects the practical application.Aiming at this problem,we proposed an intrusion classification algorithm for α core-set multilevel aggregation clustering,which is based on rough set with entropy important measurement weight.First,based on entropy important measurement weight method it uses the rough set to carry out pretreatment and attribute reduction on intrusion detection data,and to decrease data dimension for preventing the algorithm from falling into "dimension trap";Secondly,it replaces the Euclidean distance of alpha multilevel aggregation clustering algorithm with entropy important measurement weight distance to compute the individual similarity,and implements the effective docking of output data of rough set and alpha multilevel aggregation clustering algorithm;Finally,it was demonstrated through experiments that the proposed algorithm could more effectively do the detection and classification on KDD CUP 99 standard database.