摘要
针对入侵检测数据集中存在大量冗余信息及传统聚类算法的效果不佳,提出了结合主成分分析与属性权重模糊聚类算法的入侵检测方法。该方法分为特征提取和模糊聚类两阶段,使用主成分分析进行特征提取、消除冗余属性;将经主成分分析后得到新成分的贡献率作为聚类算法中属性的权重值,实现了基于属性权重的模糊聚类。在KDD-CUP99数据集中的实验结果表明,该方法能有效地降低检测训练时间和提高检测正确率。
In order to overcome the shortcomings that lots of redundancy information exists in intrusion detection data sets and classical clustering algorithms perform not perfectly, a new intrusion detection approach which combines the principle component analysis with feature-weighted fuzzy clustering is presented. The approach is splited into two steps,including feature extraction and fuzzy clustering. The principle component analysis is used to extract features and eliminate the redundancy attributes. The contribution proportion obtained from the former is used as the feature weight in the clustering algorithm, which forms the feature-weighted fuzzy clustering. Experiments on the data sets of KDDCUP99 show that this algorithm can obviously reduce the training time and meanwhile improve the accuracy of intrusion detection.
出处
《江苏科技大学学报（自然科学版）》
CAS
北大核心
2008年第2期67-70,共4页
Journal of Jiangsu University of Science and Technology:Natural Science Edition
基金
江苏省信息产业厅基金资助项目(2005106)
关键词
主成分分析
贡献率
模糊聚类
入侵检测
principle component analysis
contribution proportion
fuzzy clustering
intrusion detection
