QQ客服 官方微博 反馈留言

	A wavelet support vector machine-based robust interval number regression algorithm
	Zhong, Lingyan; Ren, Shijin
	2006-12
卷号	13E
页码	1169-1172
英文摘要	Most methods for interval number regression need to know the number of outliers priorly and require the proper initial parameters for neural networks, which is difficult in the presence of unknown outliers. Furthermore most outlier detection methods fail to detect outliers because of high dimensionality of samples. To tackle these problems, a wavelet support vector machine based robust interval number regression algorithm (RINRA) is presented. By changing epsilon -insensitive cost function of a new wavelet support vector regression, a data point with the highest frequency of detection and the higher average slack variable values can be viewed as outlier and removed from interval number training samples; then parameters of interval number regression model are determined by WSVM trained with the rest samples initially, and adjusted by gradient descent further to improve precision.
会议录出版者	WATAM PRESS
会议录出版地	C/O DCDIS JOURNAL, 317 KAREN PLACE, WATERLOO, ONTARIO N2L 6K8, CANADA
语种	英语
WOS研究方向	Mathematics
WOS记录号	WOS:000248840700030
内容类型	会议论文
源URL	[http://10.2.47.112/handle/2XS4QKH4/3486]
专题	上海财经大学
作者单位	1.Shanghai Univ Finance & Econ, Sch Informat Management & Engn, Shanghai 200433, Peoples R China; 2.Zhejiang Univ, Natl Lab Ind Control Technol, Hangzhou 310027, Zhejiang, Peoples R China
推荐引用方式 GB/T 7714	Zhong, Lingyan,Ren, Shijin. A wavelet support vector machine-based robust interval number regression algorithm[C]. 见:.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。