摘要： 邻域否定选择算法遍历每个自体样本,导致计算量大及匹配阶段重叠率高等问题。为此,对邻域否定选择算法和聚类技术进行研究,提出一种邻域检测器生成算法。将自体样本映射到构建好的邻域空间中进行聚类,同时对随机检测器予以耐受,训练出成熟的邻域检测器。在KDD CUP 1999数据集上的仿真结果表明,该算法可以缩短生成检测器的时间,有效解决高重叠问题,提高检测效率。

关键词: 入侵检测, 免疫, 邻域, 聚类, 检测器

Abstract: Neighborhood Negative Selection(NNS) algorithm needs to traverse the whole self-samples and leads to large amount of calculation.At the same time there are phenomena about overlap rate higher at matching stage.To address this issue,making in-depth study on NNS and clustering method,it proposes a novel neighborhood detector algorithm.The self-samples are mapped to neighborhood space and they are used to clustering.Random detectors are trained and become mature neighborhood detectors.The algorithm generates detectors by shortening the time and solving the high overlap problem.In KDD CUP 1999 data sets to evaluate the results of simulation show that,the algorithm can solve the above mentioned problems effectively and increase the detection efficiency.

Key words: intrusion detection, immunity, neighborhood, clustering, detector

中图分类号: 

• TP301.6