基于距离的不确定数据异常点检测研究

Distance-based outlier detection over uncertain data

针对实际数据存在不确定性的问题,提出了新的异常点检测方法。首先,定义了基于距离的不确定数据异常点检测概念;其次,设计了相应的不确定数据的异常点检测算法;再次,为降低算法时间复杂度,设计了剪枝策略;最后,实验分析说明了算法对不确定异常点检测的可行性与效率。

Aimed at the problem that the actual data exists uncertainty,a new method of outlier detection was proposed.First,the notions of distance-based outlier detection on uncertain data were defined.Then,an algorithm was designed to mine corresponding outliers over uncertain data.Third,a pruning algorithm was designed in order to reduce the time complexity.Finally,experimental studies illustrated that the algorithms have good efficiency in uncertain outlier detection.