摘要： 现有离群点数据挖掘算法在高维空间效率比较低，针对上述不足，从离群点对数据集有序性的影响角度出发，在界定分形离群点含义的基础上，利用分形理论将离群数据挖掘作为一个优化分割问题进行处理。采用推广的G-P算法计算数据集的多重分形广义维数，利用贪婪算法的思想设计FDOM算法用于求解离群数据挖掘优化问题。实验结果证明，该算法能有效地解决离群点检测问题。

关键词: 数据挖掘, 离群点检测, 分形理论, 多重分形

Abstract: According to the weakness that traditional outlier data mining algorithms have lower efficiency in high-dimension space, from the viewpoint of outlier affecting orderliness of data set, this paper considers outlier mining as an optimization segmentation problem by using fractal theory. Based on the defining fractal outlier, Grassberger-Procaccia(GP) algorithm is used to calculate multi-fractal and general dimension. A greedy algorithm named FDOM is designed to solve the optimization problems of outlier mining. Experimental result shows that the algrithm is feasible to solve the problems of outlier mining.

Key words: data mining, outlier detection, fractal theory, multifractal

中图分类号: 

• TP391