Model-based clustering of time-evolving networks through temporal exponential-family random graph models

Research output: Contribution to journalArticle

Abstract

Dynamic networks are a general language for describing time-evolving complex systems, and discrete time network models provide an emerging statistical technique for various applications. It is a fundamental research question to detect a set of nodes sharing similar connectivity patterns in time-evolving networks. Our work is primarily motivated by detecting groups based on interesting features of the time-evolving networks (e.g., stability). In this work, we propose a model-based clustering framework for time-evolving networks based on discrete time exponential-family random graph models, which simultaneously allows both modeling and detecting group structure. To choose the number of groups, we use the conditional likelihood to construct an effective model selection criterion. Furthermore, we propose an efficient variational expectation–maximization (EM) algorithm to find approximate maximum likelihood estimates of network parameters and mixing proportions. The power of our method is demonstrated in simulation studies and empirical applications to international trade networks and the collaboration networks of a large research university.

Original languageEnglish (US)
Article number104540
JournalJournal of Multivariate Analysis
Volume175
DOIs
StatePublished - Jan 1 2020

Fingerprint

Model-based Clustering
Exponential Family
Graph Model
Random Graphs
International trade
Maximum likelihood
Conditional Likelihood
Large scale systems
Model Selection Criteria
Discrete-time Model
Dynamic Networks
Expectation-maximization Algorithm
Maximum Likelihood Estimate
Network Model
Graph model
Clustering
Random graphs
Exponential family
Complex Systems
Sharing

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Numerical Analysis
  • Statistics, Probability and Uncertainty

Cite this

@article{266f90c83cb54218881f0ee216eb9260,
title = "Model-based clustering of time-evolving networks through temporal exponential-family random graph models",
abstract = "Dynamic networks are a general language for describing time-evolving complex systems, and discrete time network models provide an emerging statistical technique for various applications. It is a fundamental research question to detect a set of nodes sharing similar connectivity patterns in time-evolving networks. Our work is primarily motivated by detecting groups based on interesting features of the time-evolving networks (e.g., stability). In this work, we propose a model-based clustering framework for time-evolving networks based on discrete time exponential-family random graph models, which simultaneously allows both modeling and detecting group structure. To choose the number of groups, we use the conditional likelihood to construct an effective model selection criterion. Furthermore, we propose an efficient variational expectation–maximization (EM) algorithm to find approximate maximum likelihood estimates of network parameters and mixing proportions. The power of our method is demonstrated in simulation studies and empirical applications to international trade networks and the collaboration networks of a large research university.",
author = "Lee, {Kevin H.} and Lingzhou Xue and Hunter, {David Russell}",
year = "2020",
month = "1",
day = "1",
doi = "10.1016/j.jmva.2019.104540",
language = "English (US)",
volume = "175",
journal = "Journal of Multivariate Analysis",
issn = "0047-259X",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Model-based clustering of time-evolving networks through temporal exponential-family random graph models

AU - Lee, Kevin H.

AU - Xue, Lingzhou

AU - Hunter, David Russell

PY - 2020/1/1

Y1 - 2020/1/1

N2 - Dynamic networks are a general language for describing time-evolving complex systems, and discrete time network models provide an emerging statistical technique for various applications. It is a fundamental research question to detect a set of nodes sharing similar connectivity patterns in time-evolving networks. Our work is primarily motivated by detecting groups based on interesting features of the time-evolving networks (e.g., stability). In this work, we propose a model-based clustering framework for time-evolving networks based on discrete time exponential-family random graph models, which simultaneously allows both modeling and detecting group structure. To choose the number of groups, we use the conditional likelihood to construct an effective model selection criterion. Furthermore, we propose an efficient variational expectation–maximization (EM) algorithm to find approximate maximum likelihood estimates of network parameters and mixing proportions. The power of our method is demonstrated in simulation studies and empirical applications to international trade networks and the collaboration networks of a large research university.

AB - Dynamic networks are a general language for describing time-evolving complex systems, and discrete time network models provide an emerging statistical technique for various applications. It is a fundamental research question to detect a set of nodes sharing similar connectivity patterns in time-evolving networks. Our work is primarily motivated by detecting groups based on interesting features of the time-evolving networks (e.g., stability). In this work, we propose a model-based clustering framework for time-evolving networks based on discrete time exponential-family random graph models, which simultaneously allows both modeling and detecting group structure. To choose the number of groups, we use the conditional likelihood to construct an effective model selection criterion. Furthermore, we propose an efficient variational expectation–maximization (EM) algorithm to find approximate maximum likelihood estimates of network parameters and mixing proportions. The power of our method is demonstrated in simulation studies and empirical applications to international trade networks and the collaboration networks of a large research university.

UR - http://www.scopus.com/inward/record.url?scp=85072289263&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072289263&partnerID=8YFLogxK

U2 - 10.1016/j.jmva.2019.104540

DO - 10.1016/j.jmva.2019.104540

M3 - Article

AN - SCOPUS:85072289263

VL - 175

JO - Journal of Multivariate Analysis

JF - Journal of Multivariate Analysis

SN - 0047-259X

M1 - 104540

ER -