摘要
针对k-means算法事先必须获知聚类数目以及难以确定初始中心的缺点,提出了一种改进的k-means聚类算法.改进后的算法首先使用了复合形和粒子群算法来选取聚类的初始中心点,然后使用k-means算法快速收敛获取聚类结果.实验表明:把改进后的算法用于网络入侵检测系统中,可以提高不需指导的异常检测的检测率,降低误检率.
In allusion to the disadvantage of having to obtain the number of clusters of datasets in advance and the sensitivity to selecting initial clustering centers in the k-means algorithm, an improved k-means clustering algorithm based on Complex-Particle Swarm Optimization is proposed. The number of clusters of data sets and the initial clustering centers in the k-means algorithm has been assigned by the improved algorithm through the Complex-Particle Swarm Optimization. The experimental results show that the improved-algorithm can improve the DR and reduce the AR of the Unsupervised Abnormal Detection in NIDS （Network Intrusion Detection System）.
出处
《中南民族大学学报（自然科学版）》
CAS
2008年第3期75-78,共4页
Journal of South-Central University for Nationalities：Natural Science Edition
基金
国家民委自然科学基金资助项目(PMZY06004)
中南民族大学大学生科研创新基金项目(cxcy2008003y)
河池学院自然科学基金资助项目(2007B-N004)
