一种改进的基于反k近邻的流数据离群点检测算法  

An Improved Stream Data Outlier Mining Algorithm Based on Reverse k Nearest Neighbors

作　　者：呼妮 王勇[1] 

机构地区：[1]西北工业大学理学院,陕西西安710114

出　　处：《计算机与现代化》2016年第8期32-35,共4页Computer and Modernization

基　　金：西北工业大学基础研究基金资助项目(JC201273)

摘　　要：现有反k邻域的流数据离群点挖掘算法存在一些不足之处,即需要遍历每个数据对象,计算复杂度较高,稳定性较差。为了解决这些问题,本文提出一种改进的基于反k近邻的离群点检测算法OL-ORND。该算法采用细胞邻域思想,加入伪反k邻域点概念(反k邻域为空集的点对象),增加了算法的严密性,从而大大提高了算法的效率和准确率。实验表明,算法具有较好的性能。The existing stream data outliers mining algorithms based on the reverse k neighbors need to traverse each data object,so the computational complexity is higher and the stability is lower. In order to solve these problems,this paper puts forward an improved outliers detection algorithm based on reverse k nearest neighbors named OL-ORND. Using the idea of cell neighbors,adding the false k reverse neighbors object concept that does not belong to the reverse k neighborhood. So that it can improve the efficiency and accuracy of the algorithm. Through the experiment,we can see that the algorithm has good performance.

关 键 词：流数据 反k近邻 细胞邻域 离群点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]