A novel estimation of the regularization parameter for ε-SVM

E. G. Ortiz-García; J. Gascón-Moreno; S. Salcedo-Sanz; A. M. Pérez-Bellido; J. A. Portilla-Figueras; L. Carro-Calvo

doi:10.1007/978-3-642-04394-9_5

A novel estimation of the regularization parameter for ε-SVM

E. G. Ortiz-García, J. Gascón-Moreno, S. Salcedo-Sanz, A. M. Pérez-Bellido, J. A. Portilla-Figueras, L. Carro-Calvo

Chemical Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper presents a novel way of estimating the regularization parameter C in regression ε-SVM. The proposed estimation method is based on the calculation of maximum values of the generalization and error loss function terms, present in the objective function of the SVM definition. Assuming that both terms must be optimized in approximately equal conditions in the objective function, we propose to estimate C as a comparison of the new model based on maximums and the standard SVM model. The performance of our approach is shown in terms of SVM training time and test error in several regression problems from well known standard repositories.

Original language	English (US)
Title of host publication	Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings
Pages	34-41
Number of pages	8
DOIs	https://doi.org/10.1007/978-3-642-04394-9_5
State	Published - 2009
Event	10th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2009 - Burgos, Spain Duration: Sep 23 2009 → Sep 26 2009

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	5788 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	10th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2009
Country/Territory	Spain
City	Burgos
Period	09/23/09 → 09/26/09

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-642-04394-9_5

Cite this

Ortiz-García, E. G., Gascón-Moreno, J., Salcedo-Sanz, S., Pérez-Bellido, A. M., Portilla-Figueras, J. A., & Carro-Calvo, L. (2009). A novel estimation of the regularization parameter for ε-SVM. In Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings (pp. 34-41). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5788 LNCS). https://doi.org/10.1007/978-3-642-04394-9_5

Ortiz-García, E. G. ; Gascón-Moreno, J. ; Salcedo-Sanz, S. et al. / A novel estimation of the regularization parameter for ε-SVM. Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings. 2009. pp. 34-41 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{39a30a57dee043c59973fa93f635fee0,

title = "A novel estimation of the regularization parameter for ε-SVM",

abstract = "This paper presents a novel way of estimating the regularization parameter C in regression ε-SVM. The proposed estimation method is based on the calculation of maximum values of the generalization and error loss function terms, present in the objective function of the SVM definition. Assuming that both terms must be optimized in approximately equal conditions in the objective function, we propose to estimate C as a comparison of the new model based on maximums and the standard SVM model. The performance of our approach is shown in terms of SVM training time and test error in several regression problems from well known standard repositories.",

author = "Ortiz-Garc{\'i}a, {E. G.} and J. Gasc{\'o}n-Moreno and S. Salcedo-Sanz and P{\'e}rez-Bellido, {A. M.} and Portilla-Figueras, {J. A.} and L. Carro-Calvo",

year = "2009",

doi = "10.1007/978-3-642-04394-9_5",

language = "English (US)",

isbn = "3642043933",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "34--41",

booktitle = "Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings",

note = "10th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2009 ; Conference date: 23-09-2009 Through 26-09-2009",

}

Ortiz-García, EG, Gascón-Moreno, J, Salcedo-Sanz, S, Pérez-Bellido, AM, Portilla-Figueras, JA & Carro-Calvo, L 2009, A novel estimation of the regularization parameter for ε-SVM. in Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 5788 LNCS, pp. 34-41, 10th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2009, Burgos, Spain, 09/23/09. https://doi.org/10.1007/978-3-642-04394-9_5

A novel estimation of the regularization parameter for ε-SVM. / Ortiz-García, E. G.; Gascón-Moreno, J.; Salcedo-Sanz, S. et al.
Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings. 2009. p. 34-41 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 5788 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A novel estimation of the regularization parameter for ε-SVM

AU - Ortiz-García, E. G.

AU - Gascón-Moreno, J.

AU - Salcedo-Sanz, S.

AU - Pérez-Bellido, A. M.

AU - Portilla-Figueras, J. A.

AU - Carro-Calvo, L.

PY - 2009

Y1 - 2009

N2 - This paper presents a novel way of estimating the regularization parameter C in regression ε-SVM. The proposed estimation method is based on the calculation of maximum values of the generalization and error loss function terms, present in the objective function of the SVM definition. Assuming that both terms must be optimized in approximately equal conditions in the objective function, we propose to estimate C as a comparison of the new model based on maximums and the standard SVM model. The performance of our approach is shown in terms of SVM training time and test error in several regression problems from well known standard repositories.

AB - This paper presents a novel way of estimating the regularization parameter C in regression ε-SVM. The proposed estimation method is based on the calculation of maximum values of the generalization and error loss function terms, present in the objective function of the SVM definition. Assuming that both terms must be optimized in approximately equal conditions in the objective function, we propose to estimate C as a comparison of the new model based on maximums and the standard SVM model. The performance of our approach is shown in terms of SVM training time and test error in several regression problems from well known standard repositories.

UR - http://www.scopus.com/inward/record.url?scp=76249106131&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-04394-9_5

DO - 10.1007/978-3-642-04394-9_5

M3 - Conference contribution

AN - SCOPUS:76249106131

SN - 3642043933

SN - 9783642043932

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 34

EP - 41

BT - Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings

T2 - 10th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2009

Y2 - 23 September 2009 through 26 September 2009

ER -

Ortiz-García EG, Gascón-Moreno J, Salcedo-Sanz S, Pérez-Bellido AM, Portilla-Figueras JA, Carro-Calvo L. A novel estimation of the regularization parameter for ε-SVM. In Intelligent Data Engineering and Automated Learning - IDEAL 2009 - 10th International Conference, Proceedings. 2009. p. 34-41. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-04394-9_5

A novel estimation of the regularization parameter for ε-SVM

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this