Approximate polytope ensemble for one-class classification

P. Casale, O. Pujol, P. Radeva

Research output: Contribution to journalArticleAcademicpeer-review

19 Citations (Scopus)

Abstract

In this work, a new one-class classification ensemble strategy called Ap-proximate Polytope Ensemble is presented. The main contribution of the paper is threefold. First, the geometrical concept of convex hull is used to define the boundary of the target class defining the problem. Expan-sions and contractions of this geometrical structure are introduced in order to avoid over-fitting. Second, the decision whether a point belongs to the convex hull model in high dimensional spaces is approximated by means of random projections and an ensemble decision process. Finally, a tiling strategy is proposed in order to model non-convex structures. Experimental results show that the proposed strategy is significantly better than state of the art one-class classification methods on over 200 datasets.
Original languageEnglish
Pages (from-to)854-864
Number of pages11
JournalPattern Recognition
Volume47
Issue number2
DOIs
Publication statusPublished - 2014

Cite this

Casale, P. ; Pujol, O. ; Radeva, P. / Approximate polytope ensemble for one-class classification. In: Pattern Recognition. 2014 ; Vol. 47, No. 2. pp. 854-864.
@article{3356df6be6a14c8895258ebce0b7df03,
title = "Approximate polytope ensemble for one-class classification",
abstract = "In this work, a new one-class classification ensemble strategy called Ap-proximate Polytope Ensemble is presented. The main contribution of the paper is threefold. First, the geometrical concept of convex hull is used to define the boundary of the target class defining the problem. Expan-sions and contractions of this geometrical structure are introduced in order to avoid over-fitting. Second, the decision whether a point belongs to the convex hull model in high dimensional spaces is approximated by means of random projections and an ensemble decision process. Finally, a tiling strategy is proposed in order to model non-convex structures. Experimental results show that the proposed strategy is significantly better than state of the art one-class classification methods on over 200 datasets.",
author = "P. Casale and O. Pujol and P. Radeva",
year = "2014",
doi = "10.1016/j.patcog.2013.08.007",
language = "English",
volume = "47",
pages = "854--864",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier",
number = "2",

}

Approximate polytope ensemble for one-class classification. / Casale, P.; Pujol, O.; Radeva, P.

In: Pattern Recognition, Vol. 47, No. 2, 2014, p. 854-864.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Approximate polytope ensemble for one-class classification

AU - Casale, P.

AU - Pujol, O.

AU - Radeva, P.

PY - 2014

Y1 - 2014

N2 - In this work, a new one-class classification ensemble strategy called Ap-proximate Polytope Ensemble is presented. The main contribution of the paper is threefold. First, the geometrical concept of convex hull is used to define the boundary of the target class defining the problem. Expan-sions and contractions of this geometrical structure are introduced in order to avoid over-fitting. Second, the decision whether a point belongs to the convex hull model in high dimensional spaces is approximated by means of random projections and an ensemble decision process. Finally, a tiling strategy is proposed in order to model non-convex structures. Experimental results show that the proposed strategy is significantly better than state of the art one-class classification methods on over 200 datasets.

AB - In this work, a new one-class classification ensemble strategy called Ap-proximate Polytope Ensemble is presented. The main contribution of the paper is threefold. First, the geometrical concept of convex hull is used to define the boundary of the target class defining the problem. Expan-sions and contractions of this geometrical structure are introduced in order to avoid over-fitting. Second, the decision whether a point belongs to the convex hull model in high dimensional spaces is approximated by means of random projections and an ensemble decision process. Finally, a tiling strategy is proposed in order to model non-convex structures. Experimental results show that the proposed strategy is significantly better than state of the art one-class classification methods on over 200 datasets.

U2 - 10.1016/j.patcog.2013.08.007

DO - 10.1016/j.patcog.2013.08.007

M3 - Article

VL - 47

SP - 854

EP - 864

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 2

ER -