Accurate logistic variational message passing: algebraic and numerical details. Issue 1 (9th March 2017)

• Record Type: Journal Article
• Authors: Nolan, Tui H.; Wand, Matt P.
• Is Part Of: Stat. Volume 6:Issue 1(2017)
• Page Start: 102
• Page End: 112
• Publication Date: 2017-03-09
• Subjects: approximate Bayesian inference -- factor graph -- generalized additive models -- generalized linear mixed models -- mean field variational Bayes; Statistics -- Periodicals; 519.2
• DOI: 10.1002/sta4.139
• Languages: English
• ISSNs: 2049-1573

Abstract

Abstract : We provide full algebraic and numerical details required for fitting accurate logistic likelihood regression‐type models via variational message passing with factor graph fragments. Existing methodology of this type involves the Jaakkola–Jordan device, which is prone to poor accuracy. We examine two alternatives: the Saul–Jordan tilted bound device and conjugacy enforcement via multivariate normal prespecification of a key message. Both of these approaches appear in related literature. Our contributions facilitate immediate implementation within variational message passing schemes. Copyright © 2017 John Wiley & Sons, Ltd.