Network structure change point detection by posterior predictive discrepancy

Lingbin Bian; Tiangang Cui; Georgy Sofronov; Jonathan Keith

doi:10.1007/978-3-030-43465-6_5

Network structure change point detection by posterior predictive discrepancy

Lingbin Bian^*, Tiangang Cui, Georgy Sofronov, Jonathan Keith

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

1 Citation (Scopus)

Abstract

Detecting changes in network structure is important for research into systems as diverse as financial trading networks, social networks and brain connectivity. Here we present novel Bayesian methods for detecting network structure change points. We use the stochastic block model to quantify the likelihood of a network structure and develop a score we call posterior predictive discrepancy based on sliding windows to evaluate the model fitness to the data. The parameter space for this model includes unknown latent label vectors assigning network nodes to interacting communities. Monte Carlo techniques based on Gibbs sampling are used to efficiently sample the posterior distributions over this parameter space.

Original language	English
Title of host publication	Monte Carlo and Quasi-Monte Carlo Methods
Subtitle of host publication	MCQMC 2018, Rennes, France, July 1–6
Editors	Bruno Tuffin, Pierre L'Ecuyer
Place of Publication	Cham
Publisher	Springer, Springer Nature
Pages	107-123
Number of pages	17
ISBN (Electronic)	9783030434656
ISBN (Print)	9783030434649
DOIs	https://doi.org/10.1007/978-3-030-43465-6_5
Publication status	Published - 2020
Event	International Conference on Monte Carlo and Quasi-Monte Carlo Methods in Scientific Computing (13th : 2018) - Rennes, France Duration: 1 Jul 2018 → 6 Jul 2018 Conference number: 13th

Publication series

Name	Springer Proceedings in Mathematics and Statistics
Publisher	Springer
Volume	324
ISSN (Print)	2194-1009
ISSN (Electronic)	2194-1017

Conference

Conference	International Conference on Monte Carlo and Quasi-Monte Carlo Methods in Scientific Computing (13th : 2018)
Abbreviated title	MCQMC 2018
Country/Territory	France
City	Rennes
Period	1/07/18 → 6/07/18

Keywords

Bayesian inference
Networks
Sliding window
Stochastic block model
Gibbs sampling

Access to Document

10.1007/978-3-030-43465-6_5

Cite this

Bian, L., Cui, T., Sofronov, G., & Keith, J. (2020). Network structure change point detection by posterior predictive discrepancy. In B. Tuffin, & P. L'Ecuyer (Eds.), Monte Carlo and Quasi-Monte Carlo Methods: MCQMC 2018, Rennes, France, July 1–6 (pp. 107-123). (Springer Proceedings in Mathematics and Statistics; Vol. 324). Springer, Springer Nature. https://doi.org/10.1007/978-3-030-43465-6_5

@inproceedings{3f7b079463d7424fbc51682074ba89d2,

title = "Network structure change point detection by posterior predictive discrepancy",

abstract = "Detecting changes in network structure is important for research into systems as diverse as financial trading networks, social networks and brain connectivity. Here we present novel Bayesian methods for detecting network structure change points. We use the stochastic block model to quantify the likelihood of a network structure and develop a score we call posterior predictive discrepancy based on sliding windows to evaluate the model fitness to the data. The parameter space for this model includes unknown latent label vectors assigning network nodes to interacting communities. Monte Carlo techniques based on Gibbs sampling are used to efficiently sample the posterior distributions over this parameter space.",

keywords = "Bayesian inference, Networks, Sliding window, Stochastic block model, Gibbs sampling",

author = "Lingbin Bian and Tiangang Cui and Georgy Sofronov and Jonathan Keith",

year = "2020",

doi = "10.1007/978-3-030-43465-6_5",

language = "English",

isbn = "9783030434649",

series = "Springer Proceedings in Mathematics and Statistics",

publisher = "Springer, Springer Nature",

pages = "107--123",

editor = "Bruno Tuffin and Pierre L'Ecuyer",

booktitle = "Monte Carlo and Quasi-Monte Carlo Methods",

address = "United States",

note = "International Conference on Monte Carlo and Quasi-Monte Carlo Methods in Scientific Computing (13th : 2018), MCQMC 2018 ; Conference date: 01-07-2018 Through 06-07-2018",

}

Bian, L, Cui, T, Sofronov, G & Keith, J 2020, Network structure change point detection by posterior predictive discrepancy. in B Tuffin & P L'Ecuyer (eds), Monte Carlo and Quasi-Monte Carlo Methods: MCQMC 2018, Rennes, France, July 1–6. Springer Proceedings in Mathematics and Statistics, vol. 324, Springer, Springer Nature, Cham, pp. 107-123, International Conference on Monte Carlo and Quasi-Monte Carlo Methods in Scientific Computing (13th : 2018), Rennes, France, 1/07/18. https://doi.org/10.1007/978-3-030-43465-6_5

Network structure change point detection by posterior predictive discrepancy. / Bian, Lingbin; Cui, Tiangang; Sofronov, Georgy et al.
Monte Carlo and Quasi-Monte Carlo Methods: MCQMC 2018, Rennes, France, July 1–6. ed. / Bruno Tuffin; Pierre L'Ecuyer. Cham: Springer, Springer Nature, 2020. p. 107-123 (Springer Proceedings in Mathematics and Statistics; Vol. 324).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - Network structure change point detection by posterior predictive discrepancy

AU - Bian, Lingbin

AU - Cui, Tiangang

AU - Sofronov, Georgy

AU - Keith, Jonathan

N1 - Conference code: 13th

PY - 2020

Y1 - 2020

N2 - Detecting changes in network structure is important for research into systems as diverse as financial trading networks, social networks and brain connectivity. Here we present novel Bayesian methods for detecting network structure change points. We use the stochastic block model to quantify the likelihood of a network structure and develop a score we call posterior predictive discrepancy based on sliding windows to evaluate the model fitness to the data. The parameter space for this model includes unknown latent label vectors assigning network nodes to interacting communities. Monte Carlo techniques based on Gibbs sampling are used to efficiently sample the posterior distributions over this parameter space.

AB - Detecting changes in network structure is important for research into systems as diverse as financial trading networks, social networks and brain connectivity. Here we present novel Bayesian methods for detecting network structure change points. We use the stochastic block model to quantify the likelihood of a network structure and develop a score we call posterior predictive discrepancy based on sliding windows to evaluate the model fitness to the data. The parameter space for this model includes unknown latent label vectors assigning network nodes to interacting communities. Monte Carlo techniques based on Gibbs sampling are used to efficiently sample the posterior distributions over this parameter space.

KW - Bayesian inference

KW - Networks

KW - Sliding window

KW - Stochastic block model

KW - Gibbs sampling

UR - http://www.scopus.com/inward/record.url?scp=85089429781&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-43465-6_5

DO - 10.1007/978-3-030-43465-6_5

M3 - Conference proceeding contribution

SN - 9783030434649

T3 - Springer Proceedings in Mathematics and Statistics

SP - 107

EP - 123

BT - Monte Carlo and Quasi-Monte Carlo Methods

A2 - Tuffin, Bruno

A2 - L'Ecuyer, Pierre

PB - Springer, Springer Nature

CY - Cham

T2 - International Conference on Monte Carlo and Quasi-Monte Carlo Methods in Scientific Computing (13th : 2018)

Y2 - 1 July 2018 through 6 July 2018

ER -

Network structure change point detection by posterior predictive discrepancy

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this