Mean field variational bayes for continuous sparse signal shrinkage: Pitfalls and remedies

Neville, SE; Ormerod, JT; Wand, MP

Mean field variational bayes for continuous sparse signal shrinkage: Pitfalls and remedies

Neville, SE Ormerod, JT Wand, MP

Permalink

Publication Type:: Journal Article
Citation:: Electronic Journal of Statistics, 2014, 8 pp. 1113 - 1151
Issue Date:: 2014-01-01

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Published VersionAdobe PDF (1.27 MB)

View on publisher's site

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Neville, SE	en_US
dc.contributor.author	Ormerod, JT	en_US
dc.contributor.author	Wand, MP https://orcid.org/0000-0003-2555-896X	en_US
dc.date.issued	2014-01-01	en_US
dc.identifier.citation	Electronic Journal of Statistics, 2014, 8 pp. 1113 - 1151	en_US
dc.identifier.issn	1935-7524	en_US
dc.identifier.uri	http://hdl.handle.net/10453/41192
dc.description.abstract	© 2014, Institute of Mathematical Statistics. All rights received. We investigate mean field variational approximate Bayesian inference for models that use continuous distributions, Horseshoe, Negative-Exponential-Gamma and Generalized Double Pareto, for sparse signal shrinkage. Our principal finding is that the most natural, and simplest, mean field variational Bayes algorithm can perform quite poorly due to posterior dependence among auxiliary variables. More sophisticated algorithms, based on special functions, are shown to be superior. Continued fraction approximations via Lentz's Algorithm are developed to make the algorithms practical.	en_US
dc.relation.ispartof	Electronic Journal of Statistics	en_US
dc.relation.isbasedon	10.1214/14-EJS910	en_US
dc.title	Mean field variational bayes for continuous sparse signal shrinkage: Pitfalls and remedies	en_US
dc.type	Journal Article
utslib.citation.volume	8	en_US
utslib.for	0104 Statistics	en_US
pubs.embargo.period	Not known	en_US
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Science
pubs.organisational-group	/University of Technology Sydney/Faculty of Science/School of Mathematical and Physical Sciences
utslib.copyright.status	open_access
pubs.publication-status	Published	en_US
pubs.volume	8	en_US

Abstract:

© 2014, Institute of Mathematical Statistics. All rights received. We investigate mean field variational approximate Bayesian inference for models that use continuous distributions, Horseshoe, Negative-Exponential-Gamma and Generalized Double Pareto, for sparse signal shrinkage. Our principal finding is that the most natural, and simplest, mean field variational Bayes algorithm can perform quite poorly due to posterior dependence among auxiliary variables. More sophisticated algorithms, based on special functions, are shown to be superior. Continued fraction approximations via Lentz's Algorithm are developed to make the algorithms practical.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/41192