Support vector machines regression with l 1-regularizer

Hongzhi Tong; Di Rong Chen; Fenghong Yang

doi:10.1016/j.jat.2012.06.005

Support vector machines regression with l ¹-regularizer

Hongzhi Tong^*
, Di Rong Chen
, Fenghong Yang

^*Corresponding author for this work

School of Mathematical Sciences

University of International Business and Economics
Central University of Finance and Economics

Research output: Contribution to journal › Article › peer-review

10 Scopus citations

Abstract

The classical support vector machines regression (SVMR) is known as a regularized learning algorithm in reproducing kernel Hilbert spaces (RKHS) with a ε-insensitive loss function and an RKHS norm regularizer. In this paper, we study a new SVMR algorithm where the regularization term is proportional to l ¹-norm of the coefficients in the kernel ensembles. We provide an error analysis of this algorithm, an explicit learning rate is then derived under some assumptions.

Original language	English
Pages (from-to)	1331-1344
Number of pages	14
Journal	Journal of Approximation Theory
Volume	164
Issue number	10
DOIs	https://doi.org/10.1016/j.jat.2012.06.005
State	Published - 2012

Keywords

Coefficient regularization
Error decomposition
Learning rate
Reproducing kernel Hilbert spaces
Support vector machines regression

Access to Document

10.1016/j.jat.2012.06.005

Cite this

@article{0dd0119360e741a7b626fe4f61f2fe84,

title = "Support vector machines regression with l 1-regularizer",

abstract = "The classical support vector machines regression (SVMR) is known as a regularized learning algorithm in reproducing kernel Hilbert spaces (RKHS) with a ε-insensitive loss function and an RKHS norm regularizer. In this paper, we study a new SVMR algorithm where the regularization term is proportional to l 1-norm of the coefficients in the kernel ensembles. We provide an error analysis of this algorithm, an explicit learning rate is then derived under some assumptions.",

keywords = "Coefficient regularization, Error decomposition, Learning rate, Reproducing kernel Hilbert spaces, Support vector machines regression",

author = "Hongzhi Tong and Chen, \{Di Rong\} and Fenghong Yang",

year = "2012",

doi = "10.1016/j.jat.2012.06.005",

language = "英语",

volume = "164",

pages = "1331--1344",

journal = "Journal of Approximation Theory",

issn = "0021-9045",

publisher = "Academic Press Inc.",

number = "10",

}

TY - JOUR

T1 - Support vector machines regression with l 1-regularizer

AU - Tong, Hongzhi

AU - Chen, Di Rong

AU - Yang, Fenghong

PY - 2012

Y1 - 2012

N2 - The classical support vector machines regression (SVMR) is known as a regularized learning algorithm in reproducing kernel Hilbert spaces (RKHS) with a ε-insensitive loss function and an RKHS norm regularizer. In this paper, we study a new SVMR algorithm where the regularization term is proportional to l 1-norm of the coefficients in the kernel ensembles. We provide an error analysis of this algorithm, an explicit learning rate is then derived under some assumptions.

AB - The classical support vector machines regression (SVMR) is known as a regularized learning algorithm in reproducing kernel Hilbert spaces (RKHS) with a ε-insensitive loss function and an RKHS norm regularizer. In this paper, we study a new SVMR algorithm where the regularization term is proportional to l 1-norm of the coefficients in the kernel ensembles. We provide an error analysis of this algorithm, an explicit learning rate is then derived under some assumptions.

KW - Coefficient regularization

KW - Error decomposition

KW - Learning rate

KW - Reproducing kernel Hilbert spaces

KW - Support vector machines regression

UR - https://www.scopus.com/pages/publications/84864317391

U2 - 10.1016/j.jat.2012.06.005

DO - 10.1016/j.jat.2012.06.005

M3 - 文章

AN - SCOPUS:84864317391

SN - 0021-9045

VL - 164

SP - 1331

EP - 1344

JO - Journal of Approximation Theory

JF - Journal of Approximation Theory

IS - 10

ER -

Support vector machines regression with l ¹-regularizer

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this