一种基于复杂网络结构特性的模式识别方法  

A Pattern Recognition Method Based on Architectonic Characteristic about Complex Network

作　　者：杜隆胤[1] 方冰[1] 于涛[1] 郑军[1] Du Long-yin;Fang Bin;Yu Tao;Zheng Jun(College of Mathematics and Information Science,Guiyang University,Guizhou Guiyang 550003)

机构地区：[1]贵阳学院数学与信息科学学院,贵州贵阳550003

出　　处：《电子质量》2017年第5期11-14,共4页Electronics Quality

基　　金：贵州省联合基金项目:基于复杂网络社团结构的分类算法研究(编号:黔科合J字LKG[2013]44号)

摘　　要：针对训练集识别模式节点问题,利用复杂网络聚类特性,实现模式节点的识别。以"模式节点"的模式向量作为依据计算各节点间的"距离"及"相似度",然后以"模式节点"为复杂网络节点、节点间的相似度作为边构建复杂网络。利用复杂网络中节点的核心度计算待分类节点与已知类别间的相似度,完成分类识别。In this paper,we use the architectonic characteristic of complex network to solve the problem about pattern recognition.Firstly,we use the feature vector to calculate the similarity between nodes which come from training sample set.Just for this target,we cipher"distance"between nodes.Secondly,we frame a complex network,the nodes which come from training sample set used as complex network nodes and the similarities used as the weight of complex network edge.Finally,we calculate the similarity between the pattern recognition node and classes,then we can decide that what class belong the node.

关 键 词：模式识别 复杂网络 相似度 

分 类 号：O157.5[理学—数学] TP391.4[理学—基础数学]