maximum volume clustering

ISCAS OpenIR

	maximum volume clustering
	Gang Niu; Bo Dai; Lin Shang; Masashi Sugiyama
	2011
会议名称	14th International Conference on Artificial Intelligence and Statistics, AISTATS 2011
会议录名称	Journal of Machine Learning Research
页码	561-569
会议日期	April 11, 2011 - April 13, 2011
会议地点	Fort Lauderdale, FL, United states
收录类别	EI
ISSN	1532-4435
部门归属	(1) Department of Computer Science Tokyo Institute of Technology Japan; (2) State Key Laboratory for Novel Software Technology Nanjing University China; (3) NLPR/LIAMA Institute of Automation Chinese Academy of Science China
摘要	The large volume principle proposed by Vladimir Vapnik, which advocates that hypotheses lying in an equivalence class with a larger volume are more preferable, is a useful alternative to the large margin principle. In this paper, we introduce a clustering model based on the large volume principle called maximum volume clustering (MVC), and propose two algorithms to solve it approximately: a soft-label and a hard-labelMVC algorithms based on sequential quadratic programming and semi-definite programming, respectively. Our MVC model includes spectral clustering and maximum margin clustering as special cases, and is substantially more general. We also establish the finite sample stability and an error bound for soft-label MVC method. Experiments show that the proposed MVC approach compares favorably with state-of-the-art clustering algorithms. Copyright 2011 by the authors.; The large volume principle proposed by Vladimir Vapnik, which advocates that hypotheses lying in an equivalence class with a larger volume are more preferable, is a useful alternative to the large margin principle. In this paper, we introduce a clustering model based on the large volume principle called maximum volume clustering (MVC), and propose two algorithms to solve it approximately: a soft-label and a hard-labelMVC algorithms based on sequential quadratic programming and semi-definite programming, respectively. Our MVC model includes spectral clustering and maximum margin clustering as special cases, and is substantially more general. We also establish the finite sample stability and an error bound for soft-label MVC method. Experiments show that the proposed MVC approach compares favorably with state-of-the-art clustering algorithms. Copyright 2011 by the authors.
关键词	Artificial Intelligence Clustering Algorithms Equivalence Classes Sampling
主办者	Google; Microsoft; PASCAL2; IBM; NEC - Epowered by Innovation
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16311
专题	中国科学院软件研究所
推荐引用方式 GB/T 7714	Gang Niu,Bo Dai,Lin Shang,et al. maximum volume clustering[C],2011:561-569.