Novelty Detection Using Level Set Methods

Xuemei Ding, Yuhua Li, Ammar Belatreche, Liam Maguire

Research output: Contribution to journal › Article

12 Citations (Scopus)

Abstract

This paper presents a level set boundary description (LSBD) approach for novelty detection that treats the nonlinear boundary directly in the input space. The proposed approach consists of level set function (LSF) construction, boundary evolution, and termination of the training process. It employs kernel density estimation to construct the LSF of the initial boundary for the training data set. Then, a sign of the LSF-based algorithm is proposed to evolve the boundary and make it fit more tightly in the data distribution. The training process terminates when an expected fraction of rejected normal data is reached. The evolution process utilizes the signs of the LSF values at all training data points to decide whether to expand or shrink the boundary. Extensive experiments are conducted on benchmark data sets to evaluate the proposed LSBD method and compare it against four representative novelty detection methods. The experimental results demonstrate that the novelty detector modeled with the proposed LSBD can effectively detect anomalies.
LanguageEnglish
Pages576-588
JournalIEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
Volume26
Issue number3
DOIs
Publication statusPublished - Mar 2015

Fingerprint

Detectors
Experiments

Cite this

Ding, Xuemei ; Li, Yuhua ; Belatreche, Ammar ; Maguire, Liam. / Novelty Detection Using Level Set Methods. 2015 ; Vol. 26, No. 3. pp. 576-588.
@article{e358becc537b459cb9ee653cd637f7f2,
title = "Novelty Detection Using Level Set Methods",
abstract = "This paper presents a level set boundary description (LSBD) approach for novelty detection that treats the nonlinear boundary directly in the input space. The proposed approach consists of level set function (LSF) construction, boundary evolution, and termination of the training process. It employs kernel density estimation to construct the LSF of the initial boundary for the training data set. Then, a sign of the LSF-based algorithm is proposed to evolve the boundary and make it fit more tightly in the data distribution. The training process terminates when an expected fraction of rejected normal data is reached. The evolution process utilizes the signs of the LSF values at all training data points to decide whether to expand or shrink the boundary. Extensive experiments are conducted on benchmark data sets to evaluate the proposed LSBD method and compare it against four representative novelty detection methods. The experimental results demonstrate that the novelty detector modeled with the proposed LSBD can effectively detect anomalies.",
author = "Xuemei Ding and Yuhua Li and Ammar Belatreche and Liam Maguire",
year = "2015",
month = "3",
doi = "10.1109/TNNLS.2014.2320293",
language = "English",
volume = "26",
pages = "576--588",
number = "3",

}

Ding, X, Li, Y, Belatreche, A & Maguire, L 2015, 'Novelty Detection Using Level Set Methods', vol. 26, no. 3, pp. 576-588. https://doi.org/10.1109/TNNLS.2014.2320293

Novelty Detection Using Level Set Methods. / Ding, Xuemei; Li, Yuhua; Belatreche, Ammar; Maguire, Liam.

Vol. 26, No. 3, 03.2015, p. 576-588.

Research output: Contribution to journal › Article

TY - JOUR

T1 - Novelty Detection Using Level Set Methods

AU - Ding, Xuemei

AU - Li, Yuhua

AU - Belatreche, Ammar

AU - Maguire, Liam

PY - 2015/3

Y1 - 2015/3

N2 - This paper presents a level set boundary description (LSBD) approach for novelty detection that treats the nonlinear boundary directly in the input space. The proposed approach consists of level set function (LSF) construction, boundary evolution, and termination of the training process. It employs kernel density estimation to construct the LSF of the initial boundary for the training data set. Then, a sign of the LSF-based algorithm is proposed to evolve the boundary and make it fit more tightly in the data distribution. The training process terminates when an expected fraction of rejected normal data is reached. The evolution process utilizes the signs of the LSF values at all training data points to decide whether to expand or shrink the boundary. Extensive experiments are conducted on benchmark data sets to evaluate the proposed LSBD method and compare it against four representative novelty detection methods. The experimental results demonstrate that the novelty detector modeled with the proposed LSBD can effectively detect anomalies.

AB - This paper presents a level set boundary description (LSBD) approach for novelty detection that treats the nonlinear boundary directly in the input space. The proposed approach consists of level set function (LSF) construction, boundary evolution, and termination of the training process. It employs kernel density estimation to construct the LSF of the initial boundary for the training data set. Then, a sign of the LSF-based algorithm is proposed to evolve the boundary and make it fit more tightly in the data distribution. The training process terminates when an expected fraction of rejected normal data is reached. The evolution process utilizes the signs of the LSF values at all training data points to decide whether to expand or shrink the boundary. Extensive experiments are conducted on benchmark data sets to evaluate the proposed LSBD method and compare it against four representative novelty detection methods. The experimental results demonstrate that the novelty detector modeled with the proposed LSBD can effectively detect anomalies.

U2 - 10.1109/TNNLS.2014.2320293

DO - 10.1109/TNNLS.2014.2320293

M3 - Article

VL - 26

SP - 576

EP - 588

IS - 3

ER -

Ding X, Li Y, Belatreche A, Maguire L. Novelty Detection Using Level Set Methods. 2015 Mar;26(3):576-588. https://doi.org/10.1109/TNNLS.2014.2320293

Access to Document