A theoretical framework for multi-sphere support vector data description

Research output: A Conference proceeding or a Chapter in BookConference contribution

9 Citations (Scopus)

Abstract

In support vector data description (SVDD) a spherically shaped boundary around a normal data set is used to separate this set from abnormal data. The volume of this data description is minimized to reduce the chance of accepting abnormal data. However the SVDD does not guarantee that the single spherically shaped boundary can best describe the normal data set if there are some distinctive data distributions in this set. A better description is the use of multiple spheres, however there is currently no investigation available. In this paper, we propose a theoretical framework to multi-sphere SVDD in which an optimisation problem and an iterative algorithm are proposed to determine model parameters for multi-sphere SVDD to provide a better data description to the normal data set. We prove that the classification error will be reduced after each iteration in this learning process. Experimental results on 28 well-known data sets show that the proposed multi-sphere SVDD provides lower classification error rate comparing with the standard single-sphere SVDD
Original languageEnglish
Title of host publicationInternational Conference on Neural Information Processing, ICONIP 2010
Subtitle of host publicationNeural Information Processing. Models and Applications
Place of PublicationBerlin, Germany
PublisherSpringer
Pages132-142
Number of pages11
Volume6444
ISBN (Print)9783642175336
DOIs
Publication statusPublished - 2010
EventICONIP 2010 - 17th International Conference on Neural Information Processing - Sydney, Australia
Duration: 22 Nov 201025 Nov 2010

Conference

ConferenceICONIP 2010 - 17th International Conference on Neural Information Processing
CountryAustralia
CitySydney
Period22/11/1025/11/10

Fingerprint

Data description

Cite this

Tran, D., Ma, W., & Sharma, D. (2010). A theoretical framework for multi-sphere support vector data description. In International Conference on Neural Information Processing, ICONIP 2010: Neural Information Processing. Models and Applications (Vol. 6444, pp. 132-142). Berlin, Germany: Springer. https://doi.org/10.1007/978-3-642-17534-3_17
Tran, Dat ; Ma, Wanli ; Sharma, Dharmendra. / A theoretical framework for multi-sphere support vector data description. International Conference on Neural Information Processing, ICONIP 2010: Neural Information Processing. Models and Applications. Vol. 6444 Berlin, Germany : Springer, 2010. pp. 132-142
@inproceedings{d88f498a1d764b7598c53baa19873c3d,
title = "A theoretical framework for multi-sphere support vector data description",
abstract = "In support vector data description (SVDD) a spherically shaped boundary around a normal data set is used to separate this set from abnormal data. The volume of this data description is minimized to reduce the chance of accepting abnormal data. However the SVDD does not guarantee that the single spherically shaped boundary can best describe the normal data set if there are some distinctive data distributions in this set. A better description is the use of multiple spheres, however there is currently no investigation available. In this paper, we propose a theoretical framework to multi-sphere SVDD in which an optimisation problem and an iterative algorithm are proposed to determine model parameters for multi-sphere SVDD to provide a better data description to the normal data set. We prove that the classification error will be reduced after each iteration in this learning process. Experimental results on 28 well-known data sets show that the proposed multi-sphere SVDD provides lower classification error rate comparing with the standard single-sphere SVDD",
author = "Dat Tran and Wanli Ma and Dharmendra Sharma",
year = "2010",
doi = "10.1007/978-3-642-17534-3_17",
language = "English",
isbn = "9783642175336",
volume = "6444",
pages = "132--142",
booktitle = "International Conference on Neural Information Processing, ICONIP 2010",
publisher = "Springer",
address = "Netherlands",

}

Tran, D, Ma, W & Sharma, D 2010, A theoretical framework for multi-sphere support vector data description. in International Conference on Neural Information Processing, ICONIP 2010: Neural Information Processing. Models and Applications. vol. 6444, Springer, Berlin, Germany, pp. 132-142, ICONIP 2010 - 17th International Conference on Neural Information Processing, Sydney, Australia, 22/11/10. https://doi.org/10.1007/978-3-642-17534-3_17

A theoretical framework for multi-sphere support vector data description. / Tran, Dat; Ma, Wanli; Sharma, Dharmendra.

International Conference on Neural Information Processing, ICONIP 2010: Neural Information Processing. Models and Applications. Vol. 6444 Berlin, Germany : Springer, 2010. p. 132-142.

Research output: A Conference proceeding or a Chapter in BookConference contribution

TY - GEN

T1 - A theoretical framework for multi-sphere support vector data description

AU - Tran, Dat

AU - Ma, Wanli

AU - Sharma, Dharmendra

PY - 2010

Y1 - 2010

N2 - In support vector data description (SVDD) a spherically shaped boundary around a normal data set is used to separate this set from abnormal data. The volume of this data description is minimized to reduce the chance of accepting abnormal data. However the SVDD does not guarantee that the single spherically shaped boundary can best describe the normal data set if there are some distinctive data distributions in this set. A better description is the use of multiple spheres, however there is currently no investigation available. In this paper, we propose a theoretical framework to multi-sphere SVDD in which an optimisation problem and an iterative algorithm are proposed to determine model parameters for multi-sphere SVDD to provide a better data description to the normal data set. We prove that the classification error will be reduced after each iteration in this learning process. Experimental results on 28 well-known data sets show that the proposed multi-sphere SVDD provides lower classification error rate comparing with the standard single-sphere SVDD

AB - In support vector data description (SVDD) a spherically shaped boundary around a normal data set is used to separate this set from abnormal data. The volume of this data description is minimized to reduce the chance of accepting abnormal data. However the SVDD does not guarantee that the single spherically shaped boundary can best describe the normal data set if there are some distinctive data distributions in this set. A better description is the use of multiple spheres, however there is currently no investigation available. In this paper, we propose a theoretical framework to multi-sphere SVDD in which an optimisation problem and an iterative algorithm are proposed to determine model parameters for multi-sphere SVDD to provide a better data description to the normal data set. We prove that the classification error will be reduced after each iteration in this learning process. Experimental results on 28 well-known data sets show that the proposed multi-sphere SVDD provides lower classification error rate comparing with the standard single-sphere SVDD

U2 - 10.1007/978-3-642-17534-3_17

DO - 10.1007/978-3-642-17534-3_17

M3 - Conference contribution

SN - 9783642175336

VL - 6444

SP - 132

EP - 142

BT - International Conference on Neural Information Processing, ICONIP 2010

PB - Springer

CY - Berlin, Germany

ER -

Tran D, Ma W, Sharma D. A theoretical framework for multi-sphere support vector data description. In International Conference on Neural Information Processing, ICONIP 2010: Neural Information Processing. Models and Applications. Vol. 6444. Berlin, Germany: Springer. 2010. p. 132-142 https://doi.org/10.1007/978-3-642-17534-3_17