Abstract: This paper proposes a new outlier detection algorithm ODRKNN based on reverse k nearest neighbors. ODRKNN counts the number of each point’s reverse k nearest neighbors to reflect its isolation degree, experimental results on synthesized and real world dataset show that ODRKNN can efficiently detect outliers and has higher efficiency than outliers detection algorithm LOF and LSC.

摘要： 提出了基于反向K近邻（RKNN）的孤立点检测算法ODRKNN。ODRKNN算法用每个数据点的反向k近邻个数来衡量该数据点的偏离程度，在综合数据集和真实数据集上的实验结果表明，该算法能有效地检测出孤立点，且算法的效率高于算法LOF和LSC的效率.