基于K-way谱聚类的背景离群点检测  

Background Outlier Detection Based on K-way Spectral Clustering

作　　者：霍莉莉[1] 薛安荣[1] 

机构地区：[1]江苏大学计算机科学与通信工程学院,江苏镇江212013

出　　处：《计算机工程》2013年第3期197-202,208,共7页Computer Engineering

基　　金：国家自然科学基金资助项目(60773049);江苏省科技型中小企业技术创新基金资助项目(BC2010172);高等学校博士学科点专项科研基金资助项目(20093227110005);江苏大学高级专业人才科研启动基金资助项目(09JDG041)

摘　　要：为提高现有背景离群点检测算法背景子图划分的准确性,提出一种基于K-way谱聚类的背景离群点检测算法。构造图模型,对其进行K-way划分,使得到的背景子图具有解释性意义,从划分后的背景子图中获得离群点。实验结果表明,该算法的H指标提高50%,VI指标降低70%,其精确度有较大提高,且没有对图的结构进行改变,不会丢失重要信息。In order to improve the background subgraph classification accuracy of existing background outlier detection algorithm, this paper proposes a background outlier detection algorithm based on K-way spectral clustering. This paper establishes the diagram model, does the K-way partition to make it have explanatory significance for background subgraph, and gets the outliers from the background subgraph. Experimental results show that the accuracy of this algorithm is improved by 50% at H index and is reduced by 70% at VI index. There is no change with the structure of graph. So it cannot produce the problem of losting important information.

关 键 词：K-way谱聚类 二分法 背景离群点 随机游走 背景子图 图划分因子 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]