基于R树的高维孤立点检测算法研究与实现  

High Dimension Outlier Detection Algorithm and Implementation Based on R-tree

作　　者：李肖 

机构地区：[1]暨南大学

出　　处：《数字通信世界》2019年第6期118-121,152,共5页Digital Communication World

摘　　要：高维孤立点检测算法已经在移动互联网、金融欺诈检测、网络入侵检测、生态系统失调、天气预报等风险控制领域得到了广泛的应用。基于距离的孤立点检测思想,结合高维数据特点,参考R树的相关操作算法,设计实现基于R树的高维孤立点检测算法。算法在剪枝策略,排序方法上较以前同种算法更有优势。在真实的数据集上的实验结果表明,该算法能在高维条件下有效地检测出孤立点,并且算法的效率高于嵌入循环算法。High dimension outlier detection algorithms have been found a wide application in the internet, financial fraud detection, network intrusions detection, ecological systems imbalance, the weather forecast and risk control. After having a reference on the distance concept in outlier detection, the characters of high dimension data and the relevant operations of R-tree. this paper designs and implements a high dimension outlier detection algorithm ODBR based on R-tree. ODBR has a better performance in cutting and soiling. Experimental results on real world dataset show that ODBR can efficiently detect outliers and has higher efficiency than the algorithm based on repetition.

关 键 词：孤立点 高维 R树索引 嵌入循环 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]