Outlier detection using ball descriptions with adjustable metric

Elzbieta Pekalska; David M J Tax; Piotr Juszczak; Elzbieta Pȩkalska; Robert P W Duin

doi:10.1007/11815921_64

Outlier detection using ball descriptions with adjustable metric

Elzbieta Pekalska, David M J Tax, Piotr Juszczak, Elzbieta Pȩkalska, Robert P W Duin

Research output: Chapter in Book/Conference proceeding › Conference contribution

Abstract

Sometimes novel or outlier data has to be detected. The outliers may indicate some interesting rare event, or they should be disregarded because they cannot be reliably processed further. In the ideal case that the objects are represented by very good features, the genuine data forms a compact cluster and a good outlier measure is the distance to the cluster center. This paper proposes three new formulations to find a good cluster center together with an optimized ℓp-distance measure. Experiments show that for some real world datasets very good classification results are obtained and that, more specifically, the ℓ1-distance is particularly suited for datasets containing discrete feature values. © Springer-Verlag Berlin Heidelberg 2006.

Original language	English
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)\|Lect. Notes Comput. Sci.
Publisher	Springer Nature
Pages	587-595
Number of pages	8
Volume	4109
ISBN (Print)	3540372369, 9783540372363
DOIs	https://doi.org/10.1007/11815921_64
Publication status	Published - 2006
Event	Joint IAPR International Workshops on Structural, Syntactic, and Statistical Pattern Recognition, SSPR 2006 and SPR 2006 - Hong Kong Duration: 1 Jul 2006 → … http://dblp.uni-trier.de/db/conf/sspr/sspr2006.html#DuinP06http://dblp.uni-trier.de/rec/bibtex/conf/sspr/DuinP06.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/sspr/DuinP06

Publication series

Name	Lecture Notes in Computer Science

Conference

Conference	Joint IAPR International Workshops on Structural, Syntactic, and Statistical Pattern Recognition, SSPR 2006 and SPR 2006
City	Hong Kong
Period	1/07/06 → …
Internet address	http://dblp.uni-trier.de/db/conf/sspr/sspr2006.html#DuinP06http://dblp.uni-trier.de/rec/bibtex/conf/sspr/DuinP06.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/sspr/DuinP06

Keywords

ℓp- ball
One-class classification
Outlier detection
Robustness

Access to Document

10.1007/11815921_64

Cite this

Pekalska, E., Tax, D. M. J., Juszczak, P., Pȩkalska, E., & Duin, R. P. W. (2006). Outlier detection using ball descriptions with adjustable metric. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci. (Vol. 4109, pp. 587-595). (Lecture Notes in Computer Science). Springer Nature. https://doi.org/10.1007/11815921_64

@inproceedings{8294c8ced4ee494bb7c1a580b95b37a7,

title = "Outlier detection using ball descriptions with adjustable metric",

abstract = "Sometimes novel or outlier data has to be detected. The outliers may indicate some interesting rare event, or they should be disregarded because they cannot be reliably processed further. In the ideal case that the objects are represented by very good features, the genuine data forms a compact cluster and a good outlier measure is the distance to the cluster center. This paper proposes three new formulations to find a good cluster center together with an optimized ℓp-distance measure. Experiments show that for some real world datasets very good classification results are obtained and that, more specifically, the ℓ1-distance is particularly suited for datasets containing discrete feature values. {\textcopyright} Springer-Verlag Berlin Heidelberg 2006.",

keywords = "ℓp- ball, One-class classification, Outlier detection, Robustness",

author = "Elzbieta Pekalska and Tax, {David M J} and Piotr Juszczak and Elzbieta Pȩkalska and Duin, {Robert P W}",

year = "2006",

doi = "10.1007/11815921_64",

language = "English",

isbn = "3540372369",

volume = "4109",

series = "Lecture Notes in Computer Science",

publisher = "Springer Nature",

pages = "587--595",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.",

address = "United States",

note = "Joint IAPR International Workshops on Structural, Syntactic, and Statistical Pattern Recognition, SSPR 2006 and SPR 2006 ; Conference date: 01-07-2006",

url = "http://dblp.uni-trier.de/db/conf/sspr/sspr2006.html#DuinP06http://dblp.uni-trier.de/rec/bibtex/conf/sspr/DuinP06.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/sspr/DuinP06",

}

Pekalska, E, Tax, DMJ, Juszczak, P, Pȩkalska, E & Duin, RPW 2006, Outlier detection using ball descriptions with adjustable metric. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. vol. 4109, Lecture Notes in Computer Science, Springer Nature, pp. 587-595, Joint IAPR International Workshops on Structural, Syntactic, and Statistical Pattern Recognition, SSPR 2006 and SPR 2006, Hong Kong, 1/07/06. https://doi.org/10.1007/11815921_64

Outlier detection using ball descriptions with adjustable metric. / Pekalska, Elzbieta; Tax, David M J; Juszczak, Piotr et al.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. Vol. 4109 Springer Nature, 2006. p. 587-595 (Lecture Notes in Computer Science).

Research output: Chapter in Book/Conference proceeding › Conference contribution

TY - GEN

T1 - Outlier detection using ball descriptions with adjustable metric

AU - Pekalska, Elzbieta

AU - Tax, David M J

AU - Juszczak, Piotr

AU - Pȩkalska, Elzbieta

AU - Duin, Robert P W

PY - 2006

Y1 - 2006

N2 - Sometimes novel or outlier data has to be detected. The outliers may indicate some interesting rare event, or they should be disregarded because they cannot be reliably processed further. In the ideal case that the objects are represented by very good features, the genuine data forms a compact cluster and a good outlier measure is the distance to the cluster center. This paper proposes three new formulations to find a good cluster center together with an optimized ℓp-distance measure. Experiments show that for some real world datasets very good classification results are obtained and that, more specifically, the ℓ1-distance is particularly suited for datasets containing discrete feature values. © Springer-Verlag Berlin Heidelberg 2006.

AB - Sometimes novel or outlier data has to be detected. The outliers may indicate some interesting rare event, or they should be disregarded because they cannot be reliably processed further. In the ideal case that the objects are represented by very good features, the genuine data forms a compact cluster and a good outlier measure is the distance to the cluster center. This paper proposes three new formulations to find a good cluster center together with an optimized ℓp-distance measure. Experiments show that for some real world datasets very good classification results are obtained and that, more specifically, the ℓ1-distance is particularly suited for datasets containing discrete feature values. © Springer-Verlag Berlin Heidelberg 2006.

KW - ℓp- ball

KW - One-class classification

KW - Outlier detection

KW - Robustness

U2 - 10.1007/11815921_64

DO - 10.1007/11815921_64

M3 - Conference contribution

SN - 3540372369

SN - 9783540372363

VL - 4109

T3 - Lecture Notes in Computer Science

SP - 587

EP - 595

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.

PB - Springer Nature

T2 - Joint IAPR International Workshops on Structural, Syntactic, and Statistical Pattern Recognition, SSPR 2006 and SPR 2006

Y2 - 1 July 2006

ER -

Pekalska E, Tax DMJ, Juszczak P, Pȩkalska E, Duin RPW. Outlier detection using ball descriptions with adjustable metric. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci.. Vol. 4109. Springer Nature. 2006. p. 587-595. (Lecture Notes in Computer Science). doi: 10.1007/11815921_64

Outlier detection using ball descriptions with adjustable metric

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this