Published on Thu Mar 07 2019

The Variational Predictive Natural Gradient

Da Tang, Rajesh Ranganath

Variational inference transforms posterior inference into parametric optimization. It enables the use of latent variable models where otherwise it would be impractical. Traditional natural gradients based on the Variational Predictive Natural Gradient fail to correct for correlations when the approximation is not the true.

0
0
0
Abstract

Variational inference transforms posterior inference into parametric optimization thereby enabling the use of latent variable models where otherwise impractical. However, variational inference can be finicky when different variational parameters control variables that are strongly correlated under the model. Traditional natural gradients based on the variational approximation fail to correct for correlations when the approximation is not the true posterior. To address this, we construct a new natural gradient called the Variational Predictive Natural Gradient (VPNG). Unlike traditional natural gradients for variational inference, this natural gradient accounts for the relationship between model parameters and variational parameters. We demonstrate the insight with a simple example as well as the empirical value on a classification task, a deep generative model of images, and probabilistic matrix factorization for recommendation.