摘要
本文介绍识别多类异常的句法模式识别方法。它将异常值分段(或分小区间)并提取特征,根据特征值确定基元,然后将异常表示成链。用已知异常的链设计最近邻分类器,通过它对未知异常进行分类和识别。文章中仅对异常识别给出了特征值的提取,链的形成以及求不同长度链之间距离的算法。
There have been a lot of methods of recognition of anomalies, for instance, cluscer analysis in geomathematical methods. All these are statistical methods, however they are more concerned with quality feature, ignoring the structural feature of the data This study introduces the methods of recognition of multiple anomalies by means o syntactic pattern recognition in that the values of anomalies are divided into sections (or small parts), with each of them featused in terms of eigen values, and the anomalies are represented by strings, then the known anomalus strings can be used to desing nearest neighbour classifiers, by which unknown anomaloies can be classified and recognized. This study gives the selection of eigen valusc, formation of strings and calculations of distances between strings of different lengths.
出处
《吉林大学学报（地球科学版）》
EI
CAS
1988年第3期351-356,共6页
Journal of Jilin University：Earth Science Edition
关键词
句法模式识别
统计方法
特征值
Syntactic Pattern recognition,Statistical method, eigen values
