基于局部线性加权的离群点检测方法  

Research of Detection of Outliers Based on Locally Linear Weighted Value

作　　者：徐雪松[1] 宋东明[1] 张谞[1] 张宏[1] 刘凤玉[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,南京210094

出　　处：《计算机科学》2008年第5期154-157,共4页Computer Science

摘　　要：为了提高高堆数据集合离群数据挖掘效率,在分析了传统的离群数据挖掘算法优点和缺点的基础上,提出了一种基于局部线性加权的离群点检测算法。该算法利用LLE算法的思想寻找样本数据的内在嵌入分布,并通过距离公式和离群点权值判别式进行权值数据判定,根据权值的大小标识出数据集中的离群点。仿真实验的结果表明了该方法能够有效地发现高维数据集中的离群点。与此同时,该算法具有参数估计简单、参数影响不大等优点。该算法为离群点检测问题的机器学习提供了一条新的途径。The data dimension reduction is the main method that can enhance the outliers mining efficiency based on higher-dimension data set. The research of detection of outliers based on locally linear weighted value is proposed after analyzing the advantages and disadvantages of the classical outlier mining algorithm in the paper. With the idea of Local Linear Embedding, the algorithm tries to find distribution of internal embedding of the samples, and determines value-weighted by combining distance formula with discretion formula of weighted outliers. Through these weighted values, the experts can identify the outliers easily. Simulation results illustrate that this algorithm is very efficient. Moreover,our method has the advantage of simple parameter estimation and low parameter sensitivity. Our method gives a new way for the solution of detection of outliers.

关 键 词：局部线性嵌入 高维数据 非线性降维 离群数据 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] TP273.22[自动化与计算机技术—计算机科学与技术]