Commute-time convolution kernels for graph clustering

Normawati A. Rahman, Edwin R. Hancock

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Commute time has proved to be a powerful attribute for clustering and characterising graph structure, and which is easily computed from the Laplacian spectrum. Moreover, commute time is robust to deletions of random edges and noisy edge weights. In this paper, we explore the idea of using convolution kernel to compare the distributions of commute time over pairs of graphs. We commence by computing the commute time distance in graphs. We then use a Gaussian convolution kernel to compare distributions. We use kernel kmeans for clustering and use kernel PCA for illustration using the COIL object recognition database.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages316-323
Number of pages8
Volume6218 LNCS
DOIs
Publication statusPublished - 2010
Externally publishedYes
Event7th Joint IAPR International Workshop on Structural, Syntactic, and Statistical Pattern Recognition, SSPR and SPR 2010 - Cesme, Izmir
Duration: 18 Aug 201020 Aug 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6218 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other7th Joint IAPR International Workshop on Structural, Syntactic, and Statistical Pattern Recognition, SSPR and SPR 2010
CityCesme, Izmir
Period18/8/1020/8/10

Fingerprint

Graph Clustering
Commute
Convolution
kernel
Object recognition
Distance in Graphs
Kernel PCA
Clustering
Laplacian Spectrum
K-means
Object Recognition
Graph in graph theory
Deletion
Attribute
Computing

Keywords

  • commute times
  • convolution kernel
  • graph kernel
  • laplacian

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Rahman, N. A., & Hancock, E. R. (2010). Commute-time convolution kernels for graph clustering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6218 LNCS, pp. 316-323). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6218 LNCS). https://doi.org/10.1007/978-3-642-14980-1_30

Commute-time convolution kernels for graph clustering. / Rahman, Normawati A.; Hancock, Edwin R.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6218 LNCS 2010. p. 316-323 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6218 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Rahman, NA & Hancock, ER 2010, Commute-time convolution kernels for graph clustering. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 6218 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6218 LNCS, pp. 316-323, 7th Joint IAPR International Workshop on Structural, Syntactic, and Statistical Pattern Recognition, SSPR and SPR 2010, Cesme, Izmir, 18/8/10. https://doi.org/10.1007/978-3-642-14980-1_30
Rahman NA, Hancock ER. Commute-time convolution kernels for graph clustering. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6218 LNCS. 2010. p. 316-323. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-14980-1_30
Rahman, Normawati A. ; Hancock, Edwin R. / Commute-time convolution kernels for graph clustering. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 6218 LNCS 2010. pp. 316-323 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f70fd84d0b4f42b29eb3e70bfd39a8d6,
title = "Commute-time convolution kernels for graph clustering",
abstract = "Commute time has proved to be a powerful attribute for clustering and characterising graph structure, and which is easily computed from the Laplacian spectrum. Moreover, commute time is robust to deletions of random edges and noisy edge weights. In this paper, we explore the idea of using convolution kernel to compare the distributions of commute time over pairs of graphs. We commence by computing the commute time distance in graphs. We then use a Gaussian convolution kernel to compare distributions. We use kernel kmeans for clustering and use kernel PCA for illustration using the COIL object recognition database.",
keywords = "commute times, convolution kernel, graph kernel, laplacian",
author = "Rahman, {Normawati A.} and Hancock, {Edwin R.}",
year = "2010",
doi = "10.1007/978-3-642-14980-1_30",
language = "English",
isbn = "3642149790",
volume = "6218 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "316--323",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Commute-time convolution kernels for graph clustering

AU - Rahman, Normawati A.

AU - Hancock, Edwin R.

PY - 2010

Y1 - 2010

N2 - Commute time has proved to be a powerful attribute for clustering and characterising graph structure, and which is easily computed from the Laplacian spectrum. Moreover, commute time is robust to deletions of random edges and noisy edge weights. In this paper, we explore the idea of using convolution kernel to compare the distributions of commute time over pairs of graphs. We commence by computing the commute time distance in graphs. We then use a Gaussian convolution kernel to compare distributions. We use kernel kmeans for clustering and use kernel PCA for illustration using the COIL object recognition database.

AB - Commute time has proved to be a powerful attribute for clustering and characterising graph structure, and which is easily computed from the Laplacian spectrum. Moreover, commute time is robust to deletions of random edges and noisy edge weights. In this paper, we explore the idea of using convolution kernel to compare the distributions of commute time over pairs of graphs. We commence by computing the commute time distance in graphs. We then use a Gaussian convolution kernel to compare distributions. We use kernel kmeans for clustering and use kernel PCA for illustration using the COIL object recognition database.

KW - commute times

KW - convolution kernel

KW - graph kernel

KW - laplacian

UR - http://www.scopus.com/inward/record.url?scp=77958495171&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77958495171&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-14980-1_30

DO - 10.1007/978-3-642-14980-1_30

M3 - Conference contribution

SN - 3642149790

SN - 9783642149795

VL - 6218 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 316

EP - 323

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -