内蒙古师范大学学报（自然科学汉文版）
內矇古師範大學學報（自然科學漢文版）
내몽고사범대학학보（자연과학한문판）
JOURNAL OF INNER MONGOLIA NORMAL UNIVERSITY(NATURAL SCIENCE EDITION)
2015年
4期
443-446
,共4页
邻域模型%入侵检测%检测率%误检率
鄰域模型%入侵檢測%檢測率%誤檢率
린역모형%입침검측%검측솔%오검솔
neighborhood model%intrusion detection%detection rate%false detection rate
根据对象邻域的分离度和耦合度确定初始聚类中心，提出一种基于邻域模型的 k-means 改进算法，并以 KDD CUP 99数据集为对象，对入侵检测进行了仿真实验。结果显示，改进后的算法在入侵检测率和误检率方面均优于 IKCM 算法和传统的 k-means 算法。
根據對象鄰域的分離度和耦閤度確定初始聚類中心，提齣一種基于鄰域模型的 k-means 改進算法，併以 KDD CUP 99數據集為對象，對入侵檢測進行瞭倣真實驗。結果顯示，改進後的算法在入侵檢測率和誤檢率方麵均優于 IKCM 算法和傳統的 k-means 算法。
근거대상린역적분리도화우합도학정초시취류중심，제출일충기우린역모형적 k-means 개진산법，병이 KDD CUP 99수거집위대상，대입침검측진행료방진실험。결과현시，개진후적산법재입침검측솔화오검솔방면균우우 IKCM 산법화전통적 k-means 산법。
This paper constructively advances the k-means,an improved algorithm based on neighbor-hood model by ascertaining the initial clustering centre according to the degree of separation and coupling in object neighborhood.It further carries the simulation experiment on intrusion detection system with the KDD CUP 99 data sets as the experimental subjects.It strongly supports the conclusion that the improved algorithm is superior to IKCM algorithm and the traditional k-means algorithm either in intrusion detection rate and false detection rate.
