ONLINE REGRESSION WITH VARYING GAUSSIANS AND NON-IDENTICAL DISTRIBUTIONS

被引：18

作者：

Hu, Ting ^{[1
]}

机构：

[1] Wuhan Univ, Sch Math & Stat, Wuhan 430072, Peoples R China

来源：

ANALYSIS AND APPLICATIONS | 2011年 / 9卷 / 04期

关键词：

Gaussian kernel; variance of Gaussian; online learning; regression algorithm; convex loss function; reproducing kernel Hilbert space; CLASSIFICATION;

D O I：

10.1142/S0219530511001923

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

We consider a fully online regression algorithm associated with a general convex loss function and Gaussian kernels with changing variances. Error analysis is conducted in a setting with samples drawn from a non-identical sequence of probability measures. When a fixed Gaussian is used, it was known that the learning ability of induced algorithms is weak. By allowing varying Gaussians, we show that the achieved learning rates can be of polynomial decays.

引用

页码：395 / 408

页数：14

共 14 条

[1]

Adams R., 1985, Sobolev Spaces

[2]

[Anonymous], ONLINE LEARNING STOC

[3] On the generalization ability of on-line learning algorithms [J].

Cesa-Bianchi, N ;

Conconi, A ;

Gentile, C .

IEEE TRANSACTIONS ON INFORMATION THEORY, 2004, 50 (09) :2050-2057

[4]

DUCHI J, 2010, P 23 ANN WORKSH COMP, P257

[5]

Hu T, 2009, J MACH LEARN RES, V10, P2873

[6] Online Classification with Varying Gaussians [J].

Hu, Ting ;

Zhou, Ding-Xuan .

STUDIES IN APPLIED MATHEMATICS, 2010, 124 (01) :65-83

[7] Online learning algorithms [J].

Smale, Steve ;

Yao, Yuan .

FOUNDATIONS OF COMPUTATIONAL MATHEMATICS, 2006, 6 (02) :145-170

[8] ESTIMATING THE APPROXIMATION ERROR IN LEARNING THEORY [J].

Smale, Steve ;

Zhou, Ding-Xuan .

ANALYSIS AND APPLICATIONS, 2003, 1 (01) :17-41

[9] ONLINE LEARNING WITH MARKOV SAMPLING [J].

Smale, Steve ;

Zhou, Ding-Xuan .

ANALYSIS AND APPLICATIONS, 2009, 7 (01) :87-113

[10] An explicit description of the reproducing kernel Hilbert spaces of Gaussian RBF kernels [J].

Steinwart, Ingo ;

Hush, Don ;

Scovel, Clint .

IEEE TRANSACTIONS ON INFORMATION THEORY, 2006, 52 (10) :4635-4643

← 1 2 →