Ellipsoidal support vector machines

Michinari Momma, kohei hatano, Hiroki Nakayama

Research output: Contribution to journalConference article

1 Citation (Scopus)

Abstract

This paper proposes the ellipsoidal SVM (e-SVM) that uses an ellipsoid center, in the version space, to approximate the Bayes point. Since SVM approximates it by a sphere center, e-SVM provides an extension to SVM for better approximation of the Bayes point. Although the idea has been mentioned before (Ruján (1997)), no work has been done for formulating and kernelizing the method. Starting from the maximum volume ellipsoid problem, we successfully formulate and kernelize it by employing relaxations. The resulting e-SVM optimization framework has much similarity to SVM; it is naturally extendable to other loss functions and other problems. A variant of the sequential minimal optimization is provided for efficient batch implementation. Moreover, we provide an online version of linear, or primal, e-SVM to be applicable for large-scale datasets.

Original languageEnglish
Pages (from-to)31-46
Number of pages16
JournalJournal of Machine Learning Research
Volume13
Publication statusPublished - Dec 1 2010
Event2nd Asian Conference on Machine Learning, ACML 2010 - Tokyo, Japan
Duration: Nov 8 2010Nov 10 2010

Fingerprint

Bayes
Ellipsoid
Support vector machines
Support Vector Machine
Optimization
Loss Function
Batch
Approximation
Similarity
Framework

All Science Journal Classification (ASJC) codes

  • Software
  • Control and Systems Engineering
  • Statistics and Probability
  • Artificial Intelligence

Cite this

Ellipsoidal support vector machines. / Momma, Michinari; hatano, kohei; Nakayama, Hiroki.

In: Journal of Machine Learning Research, Vol. 13, 01.12.2010, p. 31-46.

Research output: Contribution to journalConference article

Momma, Michinari ; hatano, kohei ; Nakayama, Hiroki. / Ellipsoidal support vector machines. In: Journal of Machine Learning Research. 2010 ; Vol. 13. pp. 31-46.
@article{3928d08f8ead44cfbe9a018f449b0f17,
title = "Ellipsoidal support vector machines",
abstract = "This paper proposes the ellipsoidal SVM (e-SVM) that uses an ellipsoid center, in the version space, to approximate the Bayes point. Since SVM approximates it by a sphere center, e-SVM provides an extension to SVM for better approximation of the Bayes point. Although the idea has been mentioned before (Ruj{\'a}n (1997)), no work has been done for formulating and kernelizing the method. Starting from the maximum volume ellipsoid problem, we successfully formulate and kernelize it by employing relaxations. The resulting e-SVM optimization framework has much similarity to SVM; it is naturally extendable to other loss functions and other problems. A variant of the sequential minimal optimization is provided for efficient batch implementation. Moreover, we provide an online version of linear, or primal, e-SVM to be applicable for large-scale datasets.",
author = "Michinari Momma and kohei hatano and Hiroki Nakayama",
year = "2010",
month = "12",
day = "1",
language = "English",
volume = "13",
pages = "31--46",
journal = "Journal of Machine Learning Research",
issn = "1532-4435",
publisher = "Microtome Publishing",

}

TY - JOUR

T1 - Ellipsoidal support vector machines

AU - Momma, Michinari

AU - hatano, kohei

AU - Nakayama, Hiroki

PY - 2010/12/1

Y1 - 2010/12/1

N2 - This paper proposes the ellipsoidal SVM (e-SVM) that uses an ellipsoid center, in the version space, to approximate the Bayes point. Since SVM approximates it by a sphere center, e-SVM provides an extension to SVM for better approximation of the Bayes point. Although the idea has been mentioned before (Ruján (1997)), no work has been done for formulating and kernelizing the method. Starting from the maximum volume ellipsoid problem, we successfully formulate and kernelize it by employing relaxations. The resulting e-SVM optimization framework has much similarity to SVM; it is naturally extendable to other loss functions and other problems. A variant of the sequential minimal optimization is provided for efficient batch implementation. Moreover, we provide an online version of linear, or primal, e-SVM to be applicable for large-scale datasets.

AB - This paper proposes the ellipsoidal SVM (e-SVM) that uses an ellipsoid center, in the version space, to approximate the Bayes point. Since SVM approximates it by a sphere center, e-SVM provides an extension to SVM for better approximation of the Bayes point. Although the idea has been mentioned before (Ruján (1997)), no work has been done for formulating and kernelizing the method. Starting from the maximum volume ellipsoid problem, we successfully formulate and kernelize it by employing relaxations. The resulting e-SVM optimization framework has much similarity to SVM; it is naturally extendable to other loss functions and other problems. A variant of the sequential minimal optimization is provided for efficient batch implementation. Moreover, we provide an online version of linear, or primal, e-SVM to be applicable for large-scale datasets.

UR - http://www.scopus.com/inward/record.url?scp=84876857032&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84876857032&partnerID=8YFLogxK

M3 - Conference article

VL - 13

SP - 31

EP - 46

JO - Journal of Machine Learning Research

JF - Journal of Machine Learning Research

SN - 1532-4435

ER -