Bootstrap AMG for spectral clustering
Abstract
Graph Laplacian is a popular tool for analyzing graphs, particularly in graph partitioning and clustering. Given a notion of similarity (via an adjacency matrix), graph clustering refers to identifying different groups such that vertices in the same group are more similar compared to vertices across different groups. Data clustering can be reformulated in terms of a graph clustering problem when the given set of data is represented as a graph, also known as similarity graph. In this context, eigenvectors of the graph Laplacian are often used to obtain a new geometric representation of the original data set that generally enhances cluster properties and improves cluster detection. Here, we apply a bootstrap algebraic multigrid (AMG) method that constructs a set of vectors associated with the graph Laplacian. These vectors, referred to as algebraically smooth ones, span a low-dimensional Euclidean space, which we use to represent the data, enabling cluster detection both in synthetic and in realistic well-clustered graphs. We show that, in the case of a good quality bootstrap AMG, the computed smooth vectors employed in the construction of the final AMG operator, which by construction is spectrally equivalent to the originally given graph Laplacian, accurately approximate the space in themore »
- Authors:
-
- National Research Council (CNR), Naples (Italy). Inst. for Applied Computing
- Univ. of Leeds (United Kingdom). School of Mathematics
- Portland State Univ., OR (United States). Dept. of Mathematics and Statistics; Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States). Center for Applied Scientific Computing\
- Publication Date:
- Research Org.:
- Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
- Sponsoring Org.:
- USDOE National Nuclear Security Administration (NNSA); European Union (EU); National Science Foundation (NSF)
- OSTI Identifier:
- 1669240
- Alternate Identifier(s):
- OSTI ID: 1504994
- Report Number(s):
- LLNL-JRNL-765277
Journal ID: ISSN 2577-7408; 955345
- Grant/Contract Number:
- AC52-07NA27344; 676629; DMS-1619640
- Resource Type:
- Accepted Manuscript
- Journal Name:
- Computational and Mathematical Methods
- Additional Journal Information:
- Journal Volume: 1; Journal Issue: 2; Journal ID: ISSN 2577-7408
- Publisher:
- Wiley
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 97 MATHEMATICS AND COMPUTING; algebraically smooth vectors; bootstrap AMG; graph Laplacian; spectral clustering
Citation Formats
D'Ambra, Pasqua, Cutillo, Luisa, and Vassilevski, Panayot S. Bootstrap AMG for spectral clustering. United States: N. p., 2019.
Web. doi:10.1002/cmm4.1020.
D'Ambra, Pasqua, Cutillo, Luisa, & Vassilevski, Panayot S. Bootstrap AMG for spectral clustering. United States. https://doi.org/10.1002/cmm4.1020
D'Ambra, Pasqua, Cutillo, Luisa, and Vassilevski, Panayot S. Wed .
"Bootstrap AMG for spectral clustering". United States. https://doi.org/10.1002/cmm4.1020. https://www.osti.gov/servlets/purl/1669240.
@article{osti_1669240,
title = {Bootstrap AMG for spectral clustering},
author = {D'Ambra, Pasqua and Cutillo, Luisa and Vassilevski, Panayot S.},
abstractNote = {Graph Laplacian is a popular tool for analyzing graphs, particularly in graph partitioning and clustering. Given a notion of similarity (via an adjacency matrix), graph clustering refers to identifying different groups such that vertices in the same group are more similar compared to vertices across different groups. Data clustering can be reformulated in terms of a graph clustering problem when the given set of data is represented as a graph, also known as similarity graph. In this context, eigenvectors of the graph Laplacian are often used to obtain a new geometric representation of the original data set that generally enhances cluster properties and improves cluster detection. Here, we apply a bootstrap algebraic multigrid (AMG) method that constructs a set of vectors associated with the graph Laplacian. These vectors, referred to as algebraically smooth ones, span a low-dimensional Euclidean space, which we use to represent the data, enabling cluster detection both in synthetic and in realistic well-clustered graphs. We show that, in the case of a good quality bootstrap AMG, the computed smooth vectors employed in the construction of the final AMG operator, which by construction is spectrally equivalent to the originally given graph Laplacian, accurately approximate the space in the lower portion of the spectrum of the preconditioned operator. Thus, our approach can be viewed as a spectral clustering technique associated with the generalized spectral problem (Laplace operator versus the final AMG operator), and hence, it can be seen as an extension of the classical spectral clustering that employs a standard eigenvalue problem.},
doi = {10.1002/cmm4.1020},
journal = {Computational and Mathematical Methods},
number = 2,
volume = 1,
place = {United States},
year = {Wed Apr 03 00:00:00 EDT 2019},
month = {Wed Apr 03 00:00:00 EDT 2019}
}
Works referenced in this record:
Finding community structure in very large networks
journal, December 2004
- Clauset, Aaron; Newman, M. E. J.; Moore, Cristopher
- Physical Review E, Vol. 70, Issue 6
The Elements of Statistical Learning
book, January 2009
- Hastie, Trevor; Tibshirani, Robert; Friedman, Jerome
- Springer Series in Statistics
Comparing clusterings—an information based distance
journal, May 2007
- Meilă, Marina
- Journal of Multivariate Analysis, Vol. 98, Issue 5
Depth-First Search and Linear Graph Algorithms
journal, June 1972
- Tarjan, Robert
- SIAM Journal on Computing, Vol. 1, Issue 2
Graph clustering
journal, August 2007
- Schaeffer, Satu Elisa
- Computer Science Review, Vol. 1, Issue 1
Bootstrap AMG
journal, January 2011
- Brandt, A.; Brannick, J.; Kahl, K.
- SIAM Journal on Scientific Computing, Vol. 33, Issue 2
Adaptive AMG with coarsening based on compatible weighted matching
journal, April 2013
- D’Ambra, Pasqua; Vassilevski, Panayot S.
- Computing and Visualization in Science, Vol. 16, Issue 2
The Elements of Statistical Learning
journal, August 2003
- Ziegel, Eric R.
- Technometrics, Vol. 45, Issue 3
Fast unfolding of communities in large networks
journal, October 2008
- Blondel, Vincent D.; Guillaume, Jean-Loup; Lambiotte, Renaud
- Journal of Statistical Mechanics: Theory and Experiment, Vol. 2008, Issue 10
Adaptive Smoothed Aggregation ($\alpha$SA)
journal, January 2004
- Brezina, M.; Falgout, R.; MacLachlan, S.
- SIAM Journal on Scientific Computing, Vol. 25, Issue 6
Equivalence between modularity optimization and maximum likelihood methods for community detection
journal, November 2016
- Newman, M. E. J.
- Physical Review E, Vol. 94, Issue 5
Laplacian matrices of graphs: a survey
journal, January 1994
- Merris, Russell
- Linear Algebra and its Applications, Vol. 197-198
BootCMatch: A Software Package for Bootstrap AMG Based on Graph Weighted Matching
journal, June 2018
- D’ambra, Pasqua; Filippone, Salvatore; Vassilevski, Panayot S.
- ACM Transactions on Mathematical Software, Vol. 44, Issue 4
The Elements of Statistical Learning
book, January 2001
- Hastie, Trevor; Friedman, Jerome; Tibshirani, Robert
- Springer Series in Statistics
Stochastic blockmodels and community structure in networks
journal, January 2011
- Karrer, Brian; Newman, M. E. J.
- Physical Review E, Vol. 83, Issue 1
The Elements of Statistical Learning
book, January 2009
- Hastie, Trevor; Tibshirani, Robert; Friedman, Jerome
- Springer Series in Statistics
Validation of community robustness
journal, April 2018
- Carissimo, Annamaria; Cutillo, Luisa; Feis, Italia De
- Computational Statistics & Data Analysis, Vol. 120
Relaxation-Based Coarsening and Multiscale Graph Organization
journal, January 2011
- Ron, Dorit; Safro, Ilya; Brandt, Achi
- Multiscale Modeling & Simulation, Vol. 9, Issue 1
Spectral methods for graph clustering – A survey
journal, June 2011
- Nascimento, Mariá C. V.; de Carvalho, André C. P. L. F.
- European Journal of Operational Research, Vol. 211, Issue 2
A tutorial on spectral clustering
journal, August 2007
- von Luxburg, Ulrike
- Statistics and Computing, Vol. 17, Issue 4
Lean Algebraic Multigrid (LAMG): Fast Graph Laplacian Linear Solver
journal, January 2012
- Livne, Oren E.; Brandt, Achi
- SIAM Journal on Scientific Computing, Vol. 34, Issue 4
Community detection in graphs
journal, February 2010
- Fortunato, Santo
- Physics Reports, Vol. 486, Issue 3-5
Algebraic Distance on Graphs
journal, January 2011
- Chen, Jie; Safro, Ilya
- SIAM Journal on Scientific Computing, Vol. 33, Issue 6
Dynamics of collective action to conserve a large common-pool resource
journal, April 2021
- Andersson, David; Bratsberg, Sigrid; Ringsmuth, Andrew K.
- Scientific Reports, Vol. 11, Issue 1
Adaptive Algebraic Multigrid
journal, January 2006
- Brezina, M.; Falgout, R.; MacLachlan, S.
- SIAM Journal on Scientific Computing, Vol. 27, Issue 4
Stochastic blockmodels and community structure in networks
text, January 2010
- Karrer, Brian; Newman, M. E. J.
- arXiv
Finding community structure in very large networks
text, January 2004
- Clauset, Aaron; Newman, M. E. J.; Moore, Cristopher
- arXiv