Discriminating graphs through spectral projections

Damien Fay, Hamed Haddadi, Steve Uhlig, Liam Kilmartin, Andrew W. Moore, Jérôme Kunegis, Marios Iliofotou

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

This paper proposes a novel non-parametric technique for clustering networks based on their structure. Many topological measures have been introduced in the literature to characterize topological properties of networks. These measures provide meaningful information about the structural properties of a network, but many networks share similar values of a given measure [1]. Furthermore, strong correlation between these measures occur on real-world graphs [2], so that using them to distinguish arbitrary graphs is difficult in practice [3]. Although a very complicated way to represent the information and the structural properties of a graph, the graph spectrum [4] is believed to be a signature of a graph [5]. A weighted form of the distribution of the graph spectrum, called the weighted spectral distribution (WSD), is proposed here as a feature vector. This feature vector may be related to actual structure in a graph and in addition may be used to form a metric between graphs; thus ideal for clustering purposes. To distinguish graphs, we propose to rely on two ways to project a weighted form of the eigenvalues of a graph into a low-dimensional space. The lower dimensional projection, turns out to nicely distinguish different classes of graphs, e.g. graphs from network topology generators [6-8], Internet application graphs [9], and dK-random graphs [10]. This technique can be used advantageously to separate graphs that would otherwise require complex sets of topological measures to be distinguished [9].

Original languageEnglish
Pages (from-to)3458-3468
Number of pages11
JournalComputer Networks
Volume55
Issue number15
DOIs
Publication statusPublished - 27 Oct 2011
Externally publishedYes

Fingerprint

Structural properties
Topology
Internet

Keywords

  • Graph metrics
  • Internet topology
  • Spectral graph theory
  • Topology generation

ASJC Scopus subject areas

  • Computer Networks and Communications

Cite this

Fay, D., Haddadi, H., Uhlig, S., Kilmartin, L., Moore, A. W., Kunegis, J., & Iliofotou, M. (2011). Discriminating graphs through spectral projections. Computer Networks, 55(15), 3458-3468. https://doi.org/10.1016/j.comnet.2011.06.024

Discriminating graphs through spectral projections. / Fay, Damien; Haddadi, Hamed; Uhlig, Steve; Kilmartin, Liam; Moore, Andrew W.; Kunegis, Jérôme; Iliofotou, Marios.

In: Computer Networks, Vol. 55, No. 15, 27.10.2011, p. 3458-3468.

Research output: Contribution to journalArticle

Fay, D, Haddadi, H, Uhlig, S, Kilmartin, L, Moore, AW, Kunegis, J & Iliofotou, M 2011, 'Discriminating graphs through spectral projections', Computer Networks, vol. 55, no. 15, pp. 3458-3468. https://doi.org/10.1016/j.comnet.2011.06.024
Fay D, Haddadi H, Uhlig S, Kilmartin L, Moore AW, Kunegis J et al. Discriminating graphs through spectral projections. Computer Networks. 2011 Oct 27;55(15):3458-3468. https://doi.org/10.1016/j.comnet.2011.06.024
Fay, Damien ; Haddadi, Hamed ; Uhlig, Steve ; Kilmartin, Liam ; Moore, Andrew W. ; Kunegis, Jérôme ; Iliofotou, Marios. / Discriminating graphs through spectral projections. In: Computer Networks. 2011 ; Vol. 55, No. 15. pp. 3458-3468.
@article{582d3ecb27dc4dac851b95a907ffe175,
title = "Discriminating graphs through spectral projections",
abstract = "This paper proposes a novel non-parametric technique for clustering networks based on their structure. Many topological measures have been introduced in the literature to characterize topological properties of networks. These measures provide meaningful information about the structural properties of a network, but many networks share similar values of a given measure [1]. Furthermore, strong correlation between these measures occur on real-world graphs [2], so that using them to distinguish arbitrary graphs is difficult in practice [3]. Although a very complicated way to represent the information and the structural properties of a graph, the graph spectrum [4] is believed to be a signature of a graph [5]. A weighted form of the distribution of the graph spectrum, called the weighted spectral distribution (WSD), is proposed here as a feature vector. This feature vector may be related to actual structure in a graph and in addition may be used to form a metric between graphs; thus ideal for clustering purposes. To distinguish graphs, we propose to rely on two ways to project a weighted form of the eigenvalues of a graph into a low-dimensional space. The lower dimensional projection, turns out to nicely distinguish different classes of graphs, e.g. graphs from network topology generators [6-8], Internet application graphs [9], and dK-random graphs [10]. This technique can be used advantageously to separate graphs that would otherwise require complex sets of topological measures to be distinguished [9].",
keywords = "Graph metrics, Internet topology, Spectral graph theory, Topology generation",
author = "Damien Fay and Hamed Haddadi and Steve Uhlig and Liam Kilmartin and Moore, {Andrew W.} and J{\'e}r{\^o}me Kunegis and Marios Iliofotou",
year = "2011",
month = "10",
day = "27",
doi = "10.1016/j.comnet.2011.06.024",
language = "English",
volume = "55",
pages = "3458--3468",
journal = "Computer Networks",
issn = "1389-1286",
publisher = "Elsevier",
number = "15",

}

TY - JOUR

T1 - Discriminating graphs through spectral projections

AU - Fay, Damien

AU - Haddadi, Hamed

AU - Uhlig, Steve

AU - Kilmartin, Liam

AU - Moore, Andrew W.

AU - Kunegis, Jérôme

AU - Iliofotou, Marios

PY - 2011/10/27

Y1 - 2011/10/27

N2 - This paper proposes a novel non-parametric technique for clustering networks based on their structure. Many topological measures have been introduced in the literature to characterize topological properties of networks. These measures provide meaningful information about the structural properties of a network, but many networks share similar values of a given measure [1]. Furthermore, strong correlation between these measures occur on real-world graphs [2], so that using them to distinguish arbitrary graphs is difficult in practice [3]. Although a very complicated way to represent the information and the structural properties of a graph, the graph spectrum [4] is believed to be a signature of a graph [5]. A weighted form of the distribution of the graph spectrum, called the weighted spectral distribution (WSD), is proposed here as a feature vector. This feature vector may be related to actual structure in a graph and in addition may be used to form a metric between graphs; thus ideal for clustering purposes. To distinguish graphs, we propose to rely on two ways to project a weighted form of the eigenvalues of a graph into a low-dimensional space. The lower dimensional projection, turns out to nicely distinguish different classes of graphs, e.g. graphs from network topology generators [6-8], Internet application graphs [9], and dK-random graphs [10]. This technique can be used advantageously to separate graphs that would otherwise require complex sets of topological measures to be distinguished [9].

AB - This paper proposes a novel non-parametric technique for clustering networks based on their structure. Many topological measures have been introduced in the literature to characterize topological properties of networks. These measures provide meaningful information about the structural properties of a network, but many networks share similar values of a given measure [1]. Furthermore, strong correlation between these measures occur on real-world graphs [2], so that using them to distinguish arbitrary graphs is difficult in practice [3]. Although a very complicated way to represent the information and the structural properties of a graph, the graph spectrum [4] is believed to be a signature of a graph [5]. A weighted form of the distribution of the graph spectrum, called the weighted spectral distribution (WSD), is proposed here as a feature vector. This feature vector may be related to actual structure in a graph and in addition may be used to form a metric between graphs; thus ideal for clustering purposes. To distinguish graphs, we propose to rely on two ways to project a weighted form of the eigenvalues of a graph into a low-dimensional space. The lower dimensional projection, turns out to nicely distinguish different classes of graphs, e.g. graphs from network topology generators [6-8], Internet application graphs [9], and dK-random graphs [10]. This technique can be used advantageously to separate graphs that would otherwise require complex sets of topological measures to be distinguished [9].

KW - Graph metrics

KW - Internet topology

KW - Spectral graph theory

KW - Topology generation

UR - http://www.scopus.com/inward/record.url?scp=80052430253&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80052430253&partnerID=8YFLogxK

U2 - 10.1016/j.comnet.2011.06.024

DO - 10.1016/j.comnet.2011.06.024

M3 - Article

AN - SCOPUS:80052430253

VL - 55

SP - 3458

EP - 3468

JO - Computer Networks

JF - Computer Networks

SN - 1389-1286

IS - 15

ER -