From outliers to prototypes: Ordering data

Stefan Harmeling, Guido Dornhege, David Tax, Frank Meinecke, Klaus Muller

Research output: Contribution to journalArticle

52 Citations (Scopus)

Abstract

We propose simple and fast methods based on nearest neighbors that order objects from high-dimensional data sets from typical points to untypical points. On the one hand, we show that these easy-to-compute orderings allow us to detect outliers (i.e. very untypical points) with a performance comparable to or better than other often much more sophisticated methods. On the other hand, we show how to use these orderings to detect prototypes (very typical points) which facilitate exploratory data analysis algorithms such as noisy nonlinear dimensionality reduction and clustering. Comprehensive experiments demonstrate the validity of our approach.

Original languageEnglish
Pages (from-to)1608-1618
Number of pages11
JournalNeurocomputing
Volume69
Issue number13-15
DOIs
Publication statusPublished - 2006 Aug 1
Externally publishedYes

Fingerprint

Cluster Analysis
Experiments
Datasets

Keywords

  • Clustering
  • Nearest neighbors
  • Noisy dimensionality reduction
  • Novelty detection
  • Ordering
  • Outlier detection

ASJC Scopus subject areas

  • Artificial Intelligence
  • Cellular and Molecular Neuroscience

Cite this

Harmeling, S., Dornhege, G., Tax, D., Meinecke, F., & Muller, K. (2006). From outliers to prototypes: Ordering data. Neurocomputing, 69(13-15), 1608-1618. https://doi.org/10.1016/j.neucom.2005.05.015

From outliers to prototypes : Ordering data. / Harmeling, Stefan; Dornhege, Guido; Tax, David; Meinecke, Frank; Muller, Klaus.

In: Neurocomputing, Vol. 69, No. 13-15, 01.08.2006, p. 1608-1618.

Research output: Contribution to journalArticle

Harmeling, S, Dornhege, G, Tax, D, Meinecke, F & Muller, K 2006, 'From outliers to prototypes: Ordering data', Neurocomputing, vol. 69, no. 13-15, pp. 1608-1618. https://doi.org/10.1016/j.neucom.2005.05.015
Harmeling S, Dornhege G, Tax D, Meinecke F, Muller K. From outliers to prototypes: Ordering data. Neurocomputing. 2006 Aug 1;69(13-15):1608-1618. https://doi.org/10.1016/j.neucom.2005.05.015
Harmeling, Stefan ; Dornhege, Guido ; Tax, David ; Meinecke, Frank ; Muller, Klaus. / From outliers to prototypes : Ordering data. In: Neurocomputing. 2006 ; Vol. 69, No. 13-15. pp. 1608-1618.
@article{a73a9ede0e3b44a2bc13099e0c2730a3,
title = "From outliers to prototypes: Ordering data",
abstract = "We propose simple and fast methods based on nearest neighbors that order objects from high-dimensional data sets from typical points to untypical points. On the one hand, we show that these easy-to-compute orderings allow us to detect outliers (i.e. very untypical points) with a performance comparable to or better than other often much more sophisticated methods. On the other hand, we show how to use these orderings to detect prototypes (very typical points) which facilitate exploratory data analysis algorithms such as noisy nonlinear dimensionality reduction and clustering. Comprehensive experiments demonstrate the validity of our approach.",
keywords = "Clustering, Nearest neighbors, Noisy dimensionality reduction, Novelty detection, Ordering, Outlier detection",
author = "Stefan Harmeling and Guido Dornhege and David Tax and Frank Meinecke and Klaus Muller",
year = "2006",
month = "8",
day = "1",
doi = "10.1016/j.neucom.2005.05.015",
language = "English",
volume = "69",
pages = "1608--1618",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",
number = "13-15",

}

TY - JOUR

T1 - From outliers to prototypes

T2 - Ordering data

AU - Harmeling, Stefan

AU - Dornhege, Guido

AU - Tax, David

AU - Meinecke, Frank

AU - Muller, Klaus

PY - 2006/8/1

Y1 - 2006/8/1

N2 - We propose simple and fast methods based on nearest neighbors that order objects from high-dimensional data sets from typical points to untypical points. On the one hand, we show that these easy-to-compute orderings allow us to detect outliers (i.e. very untypical points) with a performance comparable to or better than other often much more sophisticated methods. On the other hand, we show how to use these orderings to detect prototypes (very typical points) which facilitate exploratory data analysis algorithms such as noisy nonlinear dimensionality reduction and clustering. Comprehensive experiments demonstrate the validity of our approach.

AB - We propose simple and fast methods based on nearest neighbors that order objects from high-dimensional data sets from typical points to untypical points. On the one hand, we show that these easy-to-compute orderings allow us to detect outliers (i.e. very untypical points) with a performance comparable to or better than other often much more sophisticated methods. On the other hand, we show how to use these orderings to detect prototypes (very typical points) which facilitate exploratory data analysis algorithms such as noisy nonlinear dimensionality reduction and clustering. Comprehensive experiments demonstrate the validity of our approach.

KW - Clustering

KW - Nearest neighbors

KW - Noisy dimensionality reduction

KW - Novelty detection

KW - Ordering

KW - Outlier detection

UR - http://www.scopus.com/inward/record.url?scp=33745215847&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33745215847&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2005.05.015

DO - 10.1016/j.neucom.2005.05.015

M3 - Article

AN - SCOPUS:33745215847

VL - 69

SP - 1608

EP - 1618

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

IS - 13-15

ER -