A new fuzzy support vector machine based on the weighted margin

	A new fuzzy support vector machine based on the weighted margin
	Qing, T ; Jue, W
刊名	NEURAL PROCESSING LETTERS
	2004-12-01
卷号	20 期号:3 页码:139-150
关键词	fuzzy neural networks fuzzy pattern recognition margin maximal margin algorithm support vector machines
英文摘要	The ideas from fuzzy neural networks and support vector machine (SVM) are incorporated to make SVM classifiers perform better. The influence of the samples with high uncertainty can be decreased by employing the fuzzy membership to weigh the margin of each training vector. The linear separability, fuzzy margin, optimal hyperplane, generalization and soft fuzzy margin algorithms are discussed. A new optimization problem is obtained and SVM is then completely reformulated into a new fuzzy support vector machine (NFSVM). Moreover, the generation bound of NFSVM can be described. We also introduce the membership function in fuzzy neural networks to do some experiments. The results demonstrate that the proposed NFSVM can produce better results than regular SVM and Fuzzy Kernel Perceptron (FKP) in some real cases.
WOS标题词	Science & Technology ; Technology ; Life Sciences & Biomedicine
类目[WOS]	Computer Science, Artificial Intelligence ; Neurosciences
研究领域[WOS]	Computer Science ; Neurosciences & Neurology
关键词[WOS]	PERCEPTRON
收录类别	SCI
语种	英语
WOS记录号	WOS:000226369600001
公开日期	2015-12-24
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/9048]
专题	自动化研究所_09年以前成果
作者单位	1.New Star Res Inst Appl Technol, Hefei 230031, Peoples R China 2.Chinese Acad Sci, Inst Automat, Lab Complex Syst & Intelligence Sci, Beijing 100080, Peoples R China
推荐引用方式 GB/T 7714	Qing, T,Jue, W. A new fuzzy support vector machine based on the weighted margin[J]. NEURAL PROCESSING LETTERS,2004,20(3):139-150.
APA	Qing, T,&Jue, W.(2004).A new fuzzy support vector machine based on the weighted margin.NEURAL PROCESSING LETTERS,20(3),139-150.
MLA	Qing, T,et al."A new fuzzy support vector machine based on the weighted margin".NEURAL PROCESSING LETTERS 20.3(2004):139-150.