基于多元图结构子模式表示的模式识别方法*

针对统计模式识别和结构模式识别方法各自的局限性，提出一种基于多维数据多元图结构子模式表示（包括图形基元和特征基元表示）的模式识别方法，它体现了模式识别问题研究的一种新思维，具有鲜明的可视化特点。应用多维数据多元图表示原理实现无结构数据的结构化表示，提取出表征多元图图形的图形基元和特征基元，将对象表达为多元图结构特征子模式进行分类识别。利用UCI机器学习数据库中的Iris数据进行了分类实验，实验对比结果显示该方法具有较好的识别效果。

Pattern recognition based on structuralsub-pattern presentation of multivariate graph

This paper presented a novel method for pattern recognition based on graph and feature primitives presentation of multivariate graph. It showed a new thinking for the study of pattern recognition, which provided with visual method. Firstly,displayed the structural information of multivariate based on multivariate graphical presentation. Then extracted the graph primitives and feature primitives of multivariate graph and showed the object as structural sub-pattern of multivariate graph. Finally,classified the object based on the structural sub-pattern information of multivariate graph.Took some data experiments using the Iris dataset of the UCI repository of machine learning databases. The experiment results show the method for pattern recognition based on the structural sub-pattern information of multivariate graph is worth to expect.