Model selection for support vector machines via uniform design

Chien Ming Huang, Yuh Jye Lee, Dennis K.J. Lin, Su Yun Huang

Research output: Contribution to journalArticle

103 Citations (Scopus)

Abstract

The problem of choosing a good parameter setting for a better generalization performance in a learning task is the so-called model selection. A nested uniform design (UD) methodology is proposed for efficient, robust and automatic model selection for support vector machines (SVMs). The proposed method is applied to select the candidate set of parameter combinations and carry out a k-fold cross-validation to evaluate the generalization performance of each parameter combination. In contrast to conventional exhaustive grid search, this method can be treated as a deterministic analog of random search. It can dramatically cut down the number of parameter trials and also provide the flexibility to adjust the candidate set size under computational time constraint. The key theoretic advantage of the UD model selection over the grid search is that the UD points are "far more uniform"and "far more space filling" than lattice grid points. The better uniformity and space-filling phenomena make the UD selection scheme more efficient by avoiding wasteful function evaluations of close-by patterns. The proposed method is evaluated on different learning tasks, different data sets as well as different SVM algorithms.

Original languageEnglish (US)
Pages (from-to)335-346
Number of pages12
JournalComputational Statistics and Data Analysis
Volume52
Issue number1
DOIs
StatePublished - Sep 15 2007

Fingerprint

Uniform Design
Model Selection
Support vector machines
Support Vector Machine
Grid
Nested Design
Random Search
Function evaluation
Evaluation Function
Cross-validation
Uniformity
Design Methodology
Fold
Flexibility
Analogue
Support vector machine
Model selection
Evaluate
Generalization
Learning

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Computational Mathematics
  • Computational Theory and Mathematics
  • Applied Mathematics

Cite this

Huang, Chien Ming ; Lee, Yuh Jye ; Lin, Dennis K.J. ; Huang, Su Yun. / Model selection for support vector machines via uniform design. In: Computational Statistics and Data Analysis. 2007 ; Vol. 52, No. 1. pp. 335-346.
@article{3b4f334bd55b46b3b395e5a588515003,
title = "Model selection for support vector machines via uniform design",
abstract = "The problem of choosing a good parameter setting for a better generalization performance in a learning task is the so-called model selection. A nested uniform design (UD) methodology is proposed for efficient, robust and automatic model selection for support vector machines (SVMs). The proposed method is applied to select the candidate set of parameter combinations and carry out a k-fold cross-validation to evaluate the generalization performance of each parameter combination. In contrast to conventional exhaustive grid search, this method can be treated as a deterministic analog of random search. It can dramatically cut down the number of parameter trials and also provide the flexibility to adjust the candidate set size under computational time constraint. The key theoretic advantage of the UD model selection over the grid search is that the UD points are {"}far more uniform{"}and {"}far more space filling{"} than lattice grid points. The better uniformity and space-filling phenomena make the UD selection scheme more efficient by avoiding wasteful function evaluations of close-by patterns. The proposed method is evaluated on different learning tasks, different data sets as well as different SVM algorithms.",
author = "Huang, {Chien Ming} and Lee, {Yuh Jye} and Lin, {Dennis K.J.} and Huang, {Su Yun}",
year = "2007",
month = "9",
day = "15",
doi = "10.1016/j.csda.2007.02.013",
language = "English (US)",
volume = "52",
pages = "335--346",
journal = "Computational Statistics and Data Analysis",
issn = "0167-9473",
publisher = "Elsevier",
number = "1",

}

Model selection for support vector machines via uniform design. / Huang, Chien Ming; Lee, Yuh Jye; Lin, Dennis K.J.; Huang, Su Yun.

In: Computational Statistics and Data Analysis, Vol. 52, No. 1, 15.09.2007, p. 335-346.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Model selection for support vector machines via uniform design

AU - Huang, Chien Ming

AU - Lee, Yuh Jye

AU - Lin, Dennis K.J.

AU - Huang, Su Yun

PY - 2007/9/15

Y1 - 2007/9/15

N2 - The problem of choosing a good parameter setting for a better generalization performance in a learning task is the so-called model selection. A nested uniform design (UD) methodology is proposed for efficient, robust and automatic model selection for support vector machines (SVMs). The proposed method is applied to select the candidate set of parameter combinations and carry out a k-fold cross-validation to evaluate the generalization performance of each parameter combination. In contrast to conventional exhaustive grid search, this method can be treated as a deterministic analog of random search. It can dramatically cut down the number of parameter trials and also provide the flexibility to adjust the candidate set size under computational time constraint. The key theoretic advantage of the UD model selection over the grid search is that the UD points are "far more uniform"and "far more space filling" than lattice grid points. The better uniformity and space-filling phenomena make the UD selection scheme more efficient by avoiding wasteful function evaluations of close-by patterns. The proposed method is evaluated on different learning tasks, different data sets as well as different SVM algorithms.

AB - The problem of choosing a good parameter setting for a better generalization performance in a learning task is the so-called model selection. A nested uniform design (UD) methodology is proposed for efficient, robust and automatic model selection for support vector machines (SVMs). The proposed method is applied to select the candidate set of parameter combinations and carry out a k-fold cross-validation to evaluate the generalization performance of each parameter combination. In contrast to conventional exhaustive grid search, this method can be treated as a deterministic analog of random search. It can dramatically cut down the number of parameter trials and also provide the flexibility to adjust the candidate set size under computational time constraint. The key theoretic advantage of the UD model selection over the grid search is that the UD points are "far more uniform"and "far more space filling" than lattice grid points. The better uniformity and space-filling phenomena make the UD selection scheme more efficient by avoiding wasteful function evaluations of close-by patterns. The proposed method is evaluated on different learning tasks, different data sets as well as different SVM algorithms.

UR - http://www.scopus.com/inward/record.url?scp=34548279044&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34548279044&partnerID=8YFLogxK

U2 - 10.1016/j.csda.2007.02.013

DO - 10.1016/j.csda.2007.02.013

M3 - Article

AN - SCOPUS:34548279044

VL - 52

SP - 335

EP - 346

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

SN - 0167-9473

IS - 1

ER -