On the Evaluation of Outlier Detection and One-Class Classification Methods

Lorne Swersky; Henrique O. Marques; Jörg Sander; Ricardo J. G. B. Campello; Arthur Zimek

doi:10.1109/DSAA.2016.8

On the Evaluation of Outlier Detection and One-Class Classification Methods

Lorne Swersky, Henrique O. Marques, Jörg Sander, Ricardo J. G. B. Campello, Arthur Zimek

Department of Mathematics and Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

It has been shown that unsupervised outlier detection methods can be adapted to the one-class classification problem. In this paper, we focus on the comparison of oneclass classification algorithms with such adapted unsupervised outlier detection methods, improving on previous comparison studies in several important aspects. We study a number of one-class classification and unsupervised outlier detection methods in a rigorous experimental setup, comparing them on a large number of datasets with different characteristics, using different performance measures. Our experiments led to conclusions that do not fully agree with those of previous work.

Original language	English
Title of host publication	3rd IEEE International Conference on Data Science and Advanced Analytics : DSAA 2016
Editors	Randall Bilof
Volume	2016
Publisher	IEEE Press
Publication date	2016
Pages	1-10
ISBN (Print)	978-1-5090-5207-3
ISBN (Electronic)	978-1-5090-5206-6
DOIs	https://doi.org/10.1109/DSAA.2016.8
Publication status	Published - 2016
Event	3rd IEEE International Conference on Data Science and Advanced Analytics - Montreal, Canada Duration: 17. Oct 2016 → 19. Oct 2016 Conference number: 3

Conference

Conference	3rd IEEE International Conference on Data Science and Advanced Analytics
Number	3
Country/Territory	Canada
City	Montreal
Period	17/10/2016 → 19/10/2016

Keywords

Evaluation
Machine learning algorithms
One-class classification
Outlier detection
Predictive models
Semi-supervised learning
Unsupervised learning

Documents & Links

10.1109/DSAA.2016.8

SDU Link

Check access to the material in SDU Library's search engine

Cite this

@inproceedings{785bd4f97310495ba6ce18b6dc1a15cd,

title = "On the Evaluation of Outlier Detection and One-Class Classification Methods",

abstract = "It has been shown that unsupervised outlier detection methods can be adapted to the one-class classification problem. In this paper, we focus on the comparison of oneclass classification algorithms with such adapted unsupervised outlier detection methods, improving on previous comparison studies in several important aspects. We study a number of one-class classification and unsupervised outlier detection methods in a rigorous experimental setup, comparing them on a large number of datasets with different characteristics, using different performance measures. Our experiments led to conclusions that do not fully agree with those of previous work.",

keywords = "Evaluation, Machine learning algorithms, One-class classification, Outlier detection, Predictive models, Semi-supervised learning, Unsupervised learning",

author = "Lorne Swersky and Marques, \{Henrique O.\} and J{\"o}rg Sander and Campello, \{Ricardo J. G. B.\} and Arthur Zimek",

year = "2016",

doi = "10.1109/DSAA.2016.8",

language = "English",

isbn = "978-1-5090-5207-3",

volume = "2016",

pages = "1--10",

editor = "Randall Bilof",

booktitle = "3rd IEEE International Conference on Data Science and Advanced Analytics",

publisher = "IEEE Press",

note = "3rd IEEE International Conference on Data Science and Advanced Analytics , IEEE DSAA 2016 ; Conference date: 17-10-2016 Through 19-10-2016",

}

Swersky, L, Marques, HO, Sander, J, Campello, RJGB & Zimek, A 2016, On the Evaluation of Outlier Detection and One-Class Classification Methods. in R Bilof (ed.), 3rd IEEE International Conference on Data Science and Advanced Analytics : DSAA 2016 . vol. 2016, IEEE Press, pp. 1-10, 3rd IEEE International Conference on Data Science and Advanced Analytics , Montreal, Canada, 17/10/2016. https://doi.org/10.1109/DSAA.2016.8

On the Evaluation of Outlier Detection and One-Class Classification Methods. / Swersky, Lorne; Marques, Henrique O.; Sander, Jörg et al.
3rd IEEE International Conference on Data Science and Advanced Analytics : DSAA 2016 . ed. / Randall Bilof. Vol. 2016 IEEE Press, 2016. p. 1-10.

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

TY - GEN

T1 - On the Evaluation of Outlier Detection and One-Class Classification Methods

AU - Swersky, Lorne

AU - Marques, Henrique O.

AU - Sander, Jörg

AU - Campello, Ricardo J. G. B.

AU - Zimek, Arthur

N1 - Conference code: 3

PY - 2016

Y1 - 2016

N2 - It has been shown that unsupervised outlier detection methods can be adapted to the one-class classification problem. In this paper, we focus on the comparison of oneclass classification algorithms with such adapted unsupervised outlier detection methods, improving on previous comparison studies in several important aspects. We study a number of one-class classification and unsupervised outlier detection methods in a rigorous experimental setup, comparing them on a large number of datasets with different characteristics, using different performance measures. Our experiments led to conclusions that do not fully agree with those of previous work.

AB - It has been shown that unsupervised outlier detection methods can be adapted to the one-class classification problem. In this paper, we focus on the comparison of oneclass classification algorithms with such adapted unsupervised outlier detection methods, improving on previous comparison studies in several important aspects. We study a number of one-class classification and unsupervised outlier detection methods in a rigorous experimental setup, comparing them on a large number of datasets with different characteristics, using different performance measures. Our experiments led to conclusions that do not fully agree with those of previous work.

KW - Evaluation

KW - Machine learning algorithms

KW - One-class classification

KW - Outlier detection

KW - Predictive models

KW - Semi-supervised learning

KW - Unsupervised learning

U2 - 10.1109/DSAA.2016.8

DO - 10.1109/DSAA.2016.8

M3 - Article in proceedings

SN - 978-1-5090-5207-3

VL - 2016

SP - 1

EP - 10

BT - 3rd IEEE International Conference on Data Science and Advanced Analytics

A2 - Bilof, Randall

PB - IEEE Press

T2 - 3rd IEEE International Conference on Data Science and Advanced Analytics

Y2 - 17 October 2016 through 19 October 2016

ER -

On the Evaluation of Outlier Detection and One-Class Classification Methods

Abstract

Conference

Keywords

Documents & Links

SDU Link

Fingerprint

Cite this