Variational Bayesian inference for the Latent Position Cluster Model for network data

Files in This Item:
 File SizeFormat
DownloadVBLPCM.pdf415.81 kBAdobe PDF
Title: Variational Bayesian inference for the Latent Position Cluster Model for network data
Authors: Salter-Townshend, MichaelMurphy, Thomas Brendan
Permanent link: http://hdl.handle.net/10197/12538
Date: Jan-2013
Online since: 2021-10-11T14:29:28Z
Abstract: A number of recent approaches to modeling social networks have focussed on embedding the nodes in a latent “social space”. Nodes that are in close proximity are more likely to form links than those who are distant. This naturally accounts for reciprocal and transitive relationships which are commonly found in many network datasets. The Latent Position Cluster Model is one such model that also explicitly incorporates clustering by modeling the locations using a finite Gaussian mixture model. Observed covariates and sociality random effects may also be modeled. However, inference for the model via MCMC is cumbersome and thus scaling to large networks is a challenge. Variational Bayesian methods offer an alternative inference methodology for this problem. Sampling based MCMC is replaced by an optimization that requires many orders of magnitude fewer iterations to converge. A Variational Bayesian algorithm for the Latent Position Cluster Model is therefore developed and demonstrated.
Funding Details: Science Foundation Ireland
Type of material: Journal Article
Publisher: Elsevier
Journal: Computational Statistics & Data Analysis
Volume: 57
Issue: 1
Start page: 661
End page: 671
Copyright (published version): 2012 Elsevier
Keywords: Social network analysisVariational BayesLatent position cluster model
DOI: 10.1016/j.csda.2012.08.004
Language: en
Status of Item: Peer reviewed
This item is made available under a Creative Commons License: https://creativecommons.org/licenses/by-nc-nd/3.0/ie/
Appears in Collections:Mathematics and Statistics Research Collection
CASL Research Collection

Show full item record

Page view(s)

46
checked on Oct 17, 2021

Download(s)

12
checked on Oct 17, 2021

Google ScholarTM

Check

Altmetric


If you are a publisher or author and have copyright concerns for any item, please email research.repository@ucd.ie and the item will be withdrawn immediately. The author or person responsible for depositing the article will be contacted within one business day.