Variational Bayes for Elaborate Distributions
Wand, M.P., Ormerod, J.T., Padoan, S.A. & Fruhwirth, R.
We develop strategies for variational Bayes approximate inference for models containing elaborate distributions. Such models suffer from the difficulty that the parameter updates do not admit closed form solutions. We circumvent this problem through a combination of (a) specially tailored auxiliary variables, (b) univariate quadrature schemes and (c) finite mixture approximations of troublesome density functions. An accuracy assessment is conducted and the new methodology is illustrated in an application.Keywords: Asymmetric Laplace distribution; Auxiliary mixture sampling; Bayesian inference; Generalized Extreme Value distribution; Quadrature; Skew Normal Distribution.
This paper is available as a pdf (1296kB) file.