Theory and Algorithm for Learning with Dissimilarity Functions

doi:10.1162/neco.2008.08-06-805

CORC > 北京大学 > 信息科学技术学院

	Theory and Algorithm for Learning with Dissimilarity Functions
	Wang, Liwei ; Sugiyama, Masashi ; Yang, Cheng ; Hatano, Kohei ; Feng, Jufu
刊名	神经计算
	2009
关键词	COVARIATE SHIFT REPRESENTATION CLASSIFICATION RECOGNITION DISTANCE KERNELS IMAGES
DOI	10.1162/neco.2008.08-06-805
英文摘要	We study the problem of classification when only a dissimilarity function between objects is accessible. That is, data samples are represented not by feature vectors but in terms of their pairwise dissimilarities. We establish sufficient conditions for dissimilarity functions to allow building accurate classifiers. The theory immediately suggests a learning paradigm: construct an ensemble of simple classifiers, each depending on a pair of examples; then find a convex combination of them to achieve a large margin. We next develop a practical algorithm referred to as dissimilarity-based boosting (DBoost) for learning with dissimilarity functions under theoretical guidance. Experiments on a variety of databases demonstrate that the DBoost algorithm is promising for several dissimilarity measures widely used in practice.; Computer Science, Artificial Intelligence; SCI(E); PubMed; 3; ARTICLE; 5; 1459-1484; 21
语种	英语
内容类型	期刊论文
源URL	[http://ir.pku.edu.cn/handle/20.500.11897/291451]
专题	信息科学技术学院
推荐引用方式 GB/T 7714	Wang, Liwei,Sugiyama, Masashi,Yang, Cheng,et al. Theory and Algorithm for Learning with Dissimilarity Functions[J]. 神经计算,2009.
APA	Wang, Liwei,Sugiyama, Masashi,Yang, Cheng,Hatano, Kohei,&Feng, Jufu.(2009).Theory and Algorithm for Learning with Dissimilarity Functions.神经计算.
MLA	Wang, Liwei,et al."Theory and Algorithm for Learning with Dissimilarity Functions".神经计算 (2009).