基于人工神经网络的多维离群点检测算法

Outlier detecting algorithm for multidimensional datasets based on ANN

为了更加智能地检测离群点,克服传统离群点检测算法的机械性,提升多维数据集合离群点挖掘效率,在传统的离群数据挖掘算法的基础上,提出了一种基于人工神经网络的多维离群点检测算法。仿真实验结果表明,该算法具有对用户依赖性小、检测精度高的优点,为检测离群点提供了一种新的路径。

To detect outlier more intelligently, avoid the mechanical character of traditional algorithm of outlier detecting and improve the efficiency of outlier mining for multidimensional data sets, this paper proposes an algorithm of outlier detecting for multidimensional data sets based on ANN founded on the pros and cons of traditional algorithm of outlier mining. The simulation results indicate that the algorithm has the advantages of less dependent to users and higher accuracy. It opens up a new approach for the outlier detecting.