带虚警抑制的基于归一化残差的野值检测方法.PDFVIP

See discussions, stats, and author profiles for this publication at: /publication/290290273 Normalized residual-based outlier detection with false-alarm probability controlling Article · Dece ber 2015 DOI: 10.11999/JEIT150469 CITATIONS READS 0 27 4 authors, including: Xiaohu Ru National University of Defense Technology 11 PUBLICATIONS 13 CITATIONS SEE PROFILE All content following this page was uploaded by Xiaohu Ru on 11 May 2016. The user has requested enhance ent of the downloaded file. 第37 卷第12 期 电 子 与 信 息 学 报 Vol.37 No.12 2015 年12 月 Journal of Electronics Information Technology Dec. 2015 带虚警抑制的基于归一化残差的野值检测方法 汝小虎* 柳 征 姜文利 黄知涛 (国防科学技术大学电子科学与工程学院 长沙 410073) 摘 要：野值检测，或称异常值检测是模式识别和知识发现中一个重要的问题。以往的野值检测方法难以有效地抑 制虚警概率，针对这一问题，该文提出一种带监督情形下基于归一化残差(Normalized Residual, NR)的野值检测方 法。首先利用训练样本计算待考查模式的NR 值，其次比较NR 值与野值检测门限的相对大小，从而判断待考查模 式是否为野值。该文理论上推导了野值门限与虚警概率之间的关系表达式，以此为依据设置检测门限，可实现在少 量训练样本情况下仍能抑制虚警率的目的。计算机仿真和实测数据测试验证了所提方法在野值检测和虚警抑制方面 的优越性能。 关键词：模式识别；监督；野值检测；虚警概率；归一化残差 中图分类号：TP391.4; O235 文献标识码：A 文章编号：1009-5896(2015)12-2898-08 DOI: 10.11999/JEIT150469 Normalized Residual-based Outlier Detection with False-alarm Probability Controlling Ru Xiao-hu Liu Zheng Jiang Wen-li Huang Zhi-tao (College of Electronic Science and Engineering, National University of Defense Technology, Changsha 410073, China) Abstract: Outlier detection, also called anomaly detection, is an important issue in pattern recognition and knowledge discovery. Previous outlier detection methods