Information criteria for support vector machines

Kei Kobayashi, Fumiyasu Komaki

Research output: Contribution to journalArticle

13 Citations (Scopus)

Abstract

This paper presents kernel regularization information criterion (KRIC), which is a new criterion for tuning regularization parameters in kernel logistic regression (KLR) and support vector machines (SVMs). The main idea of the KRIC is based on the regularization information criterion (RIC). We derive an eigenvalue equation to calculate the KRIC and solve the problem. The computational cost for parameter tuning by the KRIC is reduced drastically by using the Nyström approximation. The test error rate of SVMs or KLR with the regularization parameter tuned by the KRIC is comparable with the one by the cross validation or evaluation of the evidence. The computational cost of the KRIC is significantly lower than the one of the other criteria.

Original languageEnglish
Pages (from-to)571-577
Number of pages7
JournalIEEE Transactions on Neural Networks
Volume17
Issue number3
DOIs
Publication statusPublished - 2006 May
Externally publishedYes

Fingerprint

Information Criterion
Support vector machines
Logistics
Support Vector Machine
Regularization
Tuning
kernel
Costs
Kernel Regression
Regularization Parameter
Logistic Regression
Computational Cost
Parameter Tuning
Cross-validation
Error Rate
Eigenvalue
Calculate
Evaluation
Approximation

Keywords

  • Kernel logistic regression (KLR)
  • Kernel machine
  • Parameter tuning
  • Regularization information criterion
  • Support vector machine (SVM)

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Theoretical Computer Science
  • Electrical and Electronic Engineering
  • Artificial Intelligence
  • Computational Theory and Mathematics
  • Hardware and Architecture

Cite this

Information criteria for support vector machines. / Kobayashi, Kei; Komaki, Fumiyasu.

In: IEEE Transactions on Neural Networks, Vol. 17, No. 3, 05.2006, p. 571-577.

Research output: Contribution to journalArticle

Kobayashi, Kei ; Komaki, Fumiyasu. / Information criteria for support vector machines. In: IEEE Transactions on Neural Networks. 2006 ; Vol. 17, No. 3. pp. 571-577.
@article{650e7c3de112454aa5bd7e79aae5f3bd,
title = "Information criteria for support vector machines",
abstract = "This paper presents kernel regularization information criterion (KRIC), which is a new criterion for tuning regularization parameters in kernel logistic regression (KLR) and support vector machines (SVMs). The main idea of the KRIC is based on the regularization information criterion (RIC). We derive an eigenvalue equation to calculate the KRIC and solve the problem. The computational cost for parameter tuning by the KRIC is reduced drastically by using the Nystr{\"o}m approximation. The test error rate of SVMs or KLR with the regularization parameter tuned by the KRIC is comparable with the one by the cross validation or evaluation of the evidence. The computational cost of the KRIC is significantly lower than the one of the other criteria.",
keywords = "Kernel logistic regression (KLR), Kernel machine, Parameter tuning, Regularization information criterion, Support vector machine (SVM)",
author = "Kei Kobayashi and Fumiyasu Komaki",
year = "2006",
month = "5",
doi = "10.1109/TNN.2006.873276",
language = "English",
volume = "17",
pages = "571--577",
journal = "IEEE Transactions on Neural Networks and Learning Systems",
issn = "2162-237X",
publisher = "IEEE Computational Intelligence Society",
number = "3",

}

TY - JOUR

T1 - Information criteria for support vector machines

AU - Kobayashi, Kei

AU - Komaki, Fumiyasu

PY - 2006/5

Y1 - 2006/5

N2 - This paper presents kernel regularization information criterion (KRIC), which is a new criterion for tuning regularization parameters in kernel logistic regression (KLR) and support vector machines (SVMs). The main idea of the KRIC is based on the regularization information criterion (RIC). We derive an eigenvalue equation to calculate the KRIC and solve the problem. The computational cost for parameter tuning by the KRIC is reduced drastically by using the Nyström approximation. The test error rate of SVMs or KLR with the regularization parameter tuned by the KRIC is comparable with the one by the cross validation or evaluation of the evidence. The computational cost of the KRIC is significantly lower than the one of the other criteria.

AB - This paper presents kernel regularization information criterion (KRIC), which is a new criterion for tuning regularization parameters in kernel logistic regression (KLR) and support vector machines (SVMs). The main idea of the KRIC is based on the regularization information criterion (RIC). We derive an eigenvalue equation to calculate the KRIC and solve the problem. The computational cost for parameter tuning by the KRIC is reduced drastically by using the Nyström approximation. The test error rate of SVMs or KLR with the regularization parameter tuned by the KRIC is comparable with the one by the cross validation or evaluation of the evidence. The computational cost of the KRIC is significantly lower than the one of the other criteria.

KW - Kernel logistic regression (KLR)

KW - Kernel machine

KW - Parameter tuning

KW - Regularization information criterion

KW - Support vector machine (SVM)

UR - http://www.scopus.com/inward/record.url?scp=33646530464&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646530464&partnerID=8YFLogxK

U2 - 10.1109/TNN.2006.873276

DO - 10.1109/TNN.2006.873276

M3 - Article

C2 - 16722163

AN - SCOPUS:33646530464

VL - 17

SP - 571

EP - 577

JO - IEEE Transactions on Neural Networks and Learning Systems

JF - IEEE Transactions on Neural Networks and Learning Systems

SN - 2162-237X

IS - 3

ER -