An Improved Spectral Clustering Algorithm Using Minimum Maximum Principle

摘要：

　　In this paper a novel document clustering spectral algorithm is proposed,which uses a minimum maximum principle.Firstly the low dimensional embedding of documents is attained by eigenvalue decomposition,and then a minimum maximum principle is used to get the initial seeds for k-means algorithm.Finally,K-means algorithm is performed to get the clustering results.Experimental results show that the clustering results found by this method is better than traditional clustering algorithm.

关键词： spectral clustering minimum maximum principle normalized mutual information

作者: Xu Xiu-fang Xu Sen Zhou Tian

作者单位: Yancheng Institute of Technology,College of Information Engineering,Yancheng 224051,China Science and technology on Underwater Acoustic Laboratory,Harbin Engineering University,Harbin 150001

会议类型: 国际会议

会议名称: 2012 International Applied Mechanics,Mechatronics Automation Symposium(2012应用力学,机电一体化自动化国际研讨会)(IAMMAS2012)

会议地点: 沈阳

会议语种:英文

页码: 1881-1884

在线出版日期: 2012-09-07（万方平台首次上网日期，不代表论文的发表时间）