Multiple Distribution Data Description Learning Method for Novelty Detection

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

8 Citations (Scopus)

Abstract

Current data description learning methods for novelty detection such as support vector data description and small sphere with large margin construct a spherically shaped boundary around a normal data set to separate this set from abnormal data. The volume of this sphere is minimized to reduce the chance of accepting abnormal data. However those learning methods do not guarantee that the single spherically shaped boundary can best describe the normal data set if there exist some distinctive data distributions in this set. We propose in this paper a new data description learning method that constructs a set of spherically shaped boundaries to provide a better data description to the normal data set. An optimisation problem is proposed and solving this problem results in an iterative learning algorithm to determine the set of spherically shaped boundaries. We prove that the classification error will be reduced after each iteration in our learning method. Experimental results on 23 well-known data sets show that the proposed method provides lower classification error rates
Original languageEnglish
Title of host publicationThe 2011 International Joint Conference on Neural Networks (IJCNN 2011)
EditorsAli A Minai
Place of PublicationSan Jose, USA
PublisherIEEE
Pages2321-2326
Number of pages6
Volume1
ISBN (Electronic)9781424496372
ISBN (Print)9781424496358
DOIs
Publication statusPublished - 31 Jul 2011
EventThe International Joint Conference on Neural Networks - San Jose, San Jose, United States
Duration: 31 Jul 20115 Aug 2011

Conference

ConferenceThe International Joint Conference on Neural Networks
Abbreviated titleIJCNN
CountryUnited States
CitySan Jose
Period31/07/115/08/11

Fingerprint

Data description
Learning algorithms

Cite this

Tran, D., Ma, W., & Sharma, D. (2011). Multiple Distribution Data Description Learning Method for Novelty Detection. In A. A. Minai (Ed.), The 2011 International Joint Conference on Neural Networks (IJCNN 2011) (Vol. 1, pp. 2321-2326). San Jose, USA: IEEE. https://doi.org/10.1109/IJCNN.2011.6033518
Tran, Dat ; Ma, Wanli ; Sharma, Dharmendra. / Multiple Distribution Data Description Learning Method for Novelty Detection. The 2011 International Joint Conference on Neural Networks (IJCNN 2011). editor / Ali A Minai. Vol. 1 San Jose, USA : IEEE, 2011. pp. 2321-2326
@inproceedings{ffc3f8aa1fb9451b840554cc6ef72d9d,
title = "Multiple Distribution Data Description Learning Method for Novelty Detection",
abstract = "Current data description learning methods for novelty detection such as support vector data description and small sphere with large margin construct a spherically shaped boundary around a normal data set to separate this set from abnormal data. The volume of this sphere is minimized to reduce the chance of accepting abnormal data. However those learning methods do not guarantee that the single spherically shaped boundary can best describe the normal data set if there exist some distinctive data distributions in this set. We propose in this paper a new data description learning method that constructs a set of spherically shaped boundaries to provide a better data description to the normal data set. An optimisation problem is proposed and solving this problem results in an iterative learning algorithm to determine the set of spherically shaped boundaries. We prove that the classification error will be reduced after each iteration in our learning method. Experimental results on 23 well-known data sets show that the proposed method provides lower classification error rates",
keywords = "Machine Learning, Novelty Detection",
author = "Dat Tran and Wanli Ma and Dharmendra Sharma",
year = "2011",
month = "7",
day = "31",
doi = "10.1109/IJCNN.2011.6033518",
language = "English",
isbn = "9781424496358",
volume = "1",
pages = "2321--2326",
editor = "Minai, {Ali A}",
booktitle = "The 2011 International Joint Conference on Neural Networks (IJCNN 2011)",
publisher = "IEEE",

}

Tran, D, Ma, W & Sharma, D 2011, Multiple Distribution Data Description Learning Method for Novelty Detection. in AA Minai (ed.), The 2011 International Joint Conference on Neural Networks (IJCNN 2011). vol. 1, IEEE, San Jose, USA, pp. 2321-2326, The International Joint Conference on Neural Networks, San Jose, United States, 31/07/11. https://doi.org/10.1109/IJCNN.2011.6033518

Multiple Distribution Data Description Learning Method for Novelty Detection. / Tran, Dat; Ma, Wanli; Sharma, Dharmendra.

The 2011 International Joint Conference on Neural Networks (IJCNN 2011). ed. / Ali A Minai. Vol. 1 San Jose, USA : IEEE, 2011. p. 2321-2326.

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

TY - GEN

T1 - Multiple Distribution Data Description Learning Method for Novelty Detection

AU - Tran, Dat

AU - Ma, Wanli

AU - Sharma, Dharmendra

PY - 2011/7/31

Y1 - 2011/7/31

N2 - Current data description learning methods for novelty detection such as support vector data description and small sphere with large margin construct a spherically shaped boundary around a normal data set to separate this set from abnormal data. The volume of this sphere is minimized to reduce the chance of accepting abnormal data. However those learning methods do not guarantee that the single spherically shaped boundary can best describe the normal data set if there exist some distinctive data distributions in this set. We propose in this paper a new data description learning method that constructs a set of spherically shaped boundaries to provide a better data description to the normal data set. An optimisation problem is proposed and solving this problem results in an iterative learning algorithm to determine the set of spherically shaped boundaries. We prove that the classification error will be reduced after each iteration in our learning method. Experimental results on 23 well-known data sets show that the proposed method provides lower classification error rates

AB - Current data description learning methods for novelty detection such as support vector data description and small sphere with large margin construct a spherically shaped boundary around a normal data set to separate this set from abnormal data. The volume of this sphere is minimized to reduce the chance of accepting abnormal data. However those learning methods do not guarantee that the single spherically shaped boundary can best describe the normal data set if there exist some distinctive data distributions in this set. We propose in this paper a new data description learning method that constructs a set of spherically shaped boundaries to provide a better data description to the normal data set. An optimisation problem is proposed and solving this problem results in an iterative learning algorithm to determine the set of spherically shaped boundaries. We prove that the classification error will be reduced after each iteration in our learning method. Experimental results on 23 well-known data sets show that the proposed method provides lower classification error rates

KW - Machine Learning

KW - Novelty Detection

U2 - 10.1109/IJCNN.2011.6033518

DO - 10.1109/IJCNN.2011.6033518

M3 - Conference contribution

SN - 9781424496358

VL - 1

SP - 2321

EP - 2326

BT - The 2011 International Joint Conference on Neural Networks (IJCNN 2011)

A2 - Minai, Ali A

PB - IEEE

CY - San Jose, USA

ER -

Tran D, Ma W, Sharma D. Multiple Distribution Data Description Learning Method for Novelty Detection. In Minai AA, editor, The 2011 International Joint Conference on Neural Networks (IJCNN 2011). Vol. 1. San Jose, USA: IEEE. 2011. p. 2321-2326 https://doi.org/10.1109/IJCNN.2011.6033518