Mixed Variational Inference

Abstract:

The Laplace approximation has been one of the workhorses of Bayesian inference. It often delivers good approximations in practice despite the fact that it does not strictly take into account where the volume of posterior density lies. Variational approaches avoid this issue by explicitly minimising the Kullback-Leibler divergence DKL between a postulated posterior and the true (unnormalised) logarithmic posterior. However, they rely on a closed form DKL in order to update the variational parameters. To address this, stochastic versions of variational inference have been devised that approximate the intractable DKL with a Monte Carlo average. This approximation allows calculating gradients with respect to the variational parameters. However, variational methods often postulate a factorised Gaussian approximating posterior. In doing so, they sacrifice a-posteriori correlations. In this work, we propose a method that combines the Laplace approximation with the variational approach. The advantages are that we maintain: applicability on non-conjugate models, posterior correlations and a reduced number of free variational parameters. Numerical experiments demonstrate improvement over the Laplace approximation and variational inference with factorised Gaussian posteriors.

SEEK ID: https://publications.h-its.org/publications/487

DOI: 10.1109/IJCNN.2019.8852348

Research Groups: Astroinformatics

Publication type: InCollection

Book Title: 2019 International Joint Conference on Neural Networks (IJCNN)

Publisher: IEEE

Citation: 2019 International Joint Conference on Neural Networks (IJCNN),pp.1-8,IEEE

Date Published: 1st Jul 2019

Registered Mode: by DOI

help Submitter
Citation
Gianniotis, N. (2019). Mixed Variational Inference. In 2019 International Joint Conference on Neural Networks (IJCNN). 2019 International Joint Conference on Neural Networks (IJCNN). IEEE. https://doi.org/10.1109/ijcnn.2019.8852348
Activity

Views: 4950

Created: 24th Oct 2019 at 14:05

Last updated: 5th Mar 2024 at 21:23

help Tags

This item has not yet been tagged.

help Attributions

None

Powered by
(v.1.14.2)
Copyright © 2008 - 2023 The University of Manchester and HITS gGmbH