Outlier detection in graphs: A study on the impact of multiple graph models

Guilherme Oliveira Campos, Edré Moreira, Wagner Meira, Arthur Zimek*

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

55 Downloads (Pure)

Abstract

Several previous works proposed techniques to detect outliers in graph data. Usually, some complex dataset is modeled as a graph and a technique for detecting outliers in graphs is applied. The impact of the graph model on the outlier detection capabilities of any method has been ignored. Here we assess the impact of the graph model on the outlier detection performance and the gains that may be achieved by using multiple graph models and combining the results obtained by these models. We show that assessing the similarity between graphs may be a guidance to determine effective combinations, as less similar graphs are complementary with respect to outlier information they provide and lead to better outlier detection.

Original languageEnglish
JournalComputer Science and Information Systems
Volume16
Issue number2
Pages (from-to)565-595
Number of pages31
ISSN1820-0214
DOIs
Publication statusPublished - 2019

Keywords

  • Ensemble
  • Multiple graph models
  • Outlier detection

Cite this

Campos, Guilherme Oliveira ; Moreira, Edré ; Meira, Wagner ; Zimek, Arthur. / Outlier detection in graphs : A study on the impact of multiple graph models. In: Computer Science and Information Systems. 2019 ; Vol. 16, No. 2. pp. 565-595.
@article{1842ad474e734cd7b042ff426b782c06,
title = "Outlier detection in graphs: A study on the impact of multiple graph models",
abstract = "Several previous works proposed techniques to detect outliers in graph data. Usually, some complex dataset is modeled as a graph and a technique for detecting outliers in graphs is applied. The impact of the graph model on the outlier detection capabilities of any method has been ignored. Here we assess the impact of the graph model on the outlier detection performance and the gains that may be achieved by using multiple graph models and combining the results obtained by these models. We show that assessing the similarity between graphs may be a guidance to determine effective combinations, as less similar graphs are complementary with respect to outlier information they provide and lead to better outlier detection.",
keywords = "Ensemble, Multiple graph models, Outlier detection",
author = "Campos, {Guilherme Oliveira} and Edr{\'e} Moreira and Wagner Meira and Arthur Zimek",
year = "2019",
doi = "10.2298/CSIS181001010C",
language = "English",
volume = "16",
pages = "565--595",
journal = "Computer Science and Information Systems",
issn = "1820-0214",
publisher = "Computer Science and Information Systems (COMSIS)",
number = "2",

}

Outlier detection in graphs : A study on the impact of multiple graph models. / Campos, Guilherme Oliveira; Moreira, Edré; Meira, Wagner; Zimek, Arthur.

In: Computer Science and Information Systems, Vol. 16, No. 2, 2019, p. 565-595.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Outlier detection in graphs

T2 - A study on the impact of multiple graph models

AU - Campos, Guilherme Oliveira

AU - Moreira, Edré

AU - Meira, Wagner

AU - Zimek, Arthur

PY - 2019

Y1 - 2019

N2 - Several previous works proposed techniques to detect outliers in graph data. Usually, some complex dataset is modeled as a graph and a technique for detecting outliers in graphs is applied. The impact of the graph model on the outlier detection capabilities of any method has been ignored. Here we assess the impact of the graph model on the outlier detection performance and the gains that may be achieved by using multiple graph models and combining the results obtained by these models. We show that assessing the similarity between graphs may be a guidance to determine effective combinations, as less similar graphs are complementary with respect to outlier information they provide and lead to better outlier detection.

AB - Several previous works proposed techniques to detect outliers in graph data. Usually, some complex dataset is modeled as a graph and a technique for detecting outliers in graphs is applied. The impact of the graph model on the outlier detection capabilities of any method has been ignored. Here we assess the impact of the graph model on the outlier detection performance and the gains that may be achieved by using multiple graph models and combining the results obtained by these models. We show that assessing the similarity between graphs may be a guidance to determine effective combinations, as less similar graphs are complementary with respect to outlier information they provide and lead to better outlier detection.

KW - Ensemble

KW - Multiple graph models

KW - Outlier detection

U2 - 10.2298/CSIS181001010C

DO - 10.2298/CSIS181001010C

M3 - Journal article

AN - SCOPUS:85071291304

VL - 16

SP - 565

EP - 595

JO - Computer Science and Information Systems

JF - Computer Science and Information Systems

SN - 1820-0214

IS - 2

ER -