Published on Mon Oct 05 2020

Unbiased Gradient Estimation for Variational Auto-Encoders using Coupled Markov Chains

Francisco J. R. Ruiz, Michalis K. Titsias, Taylan Cemgil, Arnaud Doucet

The variational auto-encoder (VAE) is a deep latent variable model that has two neural networks in an autoencoder-like architecture. Fitting its parameters via maximum likelihood (ML) is challenging since the computation of the marginal likelihood involves an intractable integral.

0
0
0
Abstract

The variational auto-encoder (VAE) is a deep latent variable model that has two neural networks in an autoencoder-like architecture; one of them parameterizes the model's likelihood. Fitting its parameters via maximum likelihood (ML) is challenging since the computation of the marginal likelihood involves an intractable integral over the latent space; thus the VAE is trained instead by maximizing a variational lower bound. Here, we develop a ML training scheme for VAEs by introducing unbiased estimators of the log-likelihood gradient. We obtain the estimators by augmenting the latent space with a set of importance samples, similarly to the importance weighted auto-encoder (IWAE), and then constructing a Markov chain Monte Carlo coupling procedure on this augmented space. We provide the conditions under which the estimators can be computed in finite time and with finite variance. We show experimentally that VAEs fitted with unbiased estimators exhibit better predictive performance.

Tue May 29 2018
Machine Learning
Hamiltonian Variational Auto-Encoder
Variational Auto-Encoders (VAEs) have become very popular techniques to perform inference and learning in latent variable models. For this methodology to be practically efficient, it is necessary to obtain low-variance unbiased estimators of the ELBO and its gradients. We show here how to optimally select reverse kernels to achieve this.
0
0
0
Fri May 20 2016
Machine Learning
Stick-Breaking Variational Autoencoders
We extend Stochastic Gradient Variational Bayes to perform posterior inference for the weights of Stick-Breaking processes. This development allows us to define a Stick-breaking Variational Autoencoder (SB-VAE)
0
0
0
Fri May 10 2019
Machine Learning
A Contrastive Divergence for Combining Variational Inference and MCMC
We develop a method to combine Markov chain Monte Carlo (MCMC) and variational inference (VI) The VCD captures a notion of discrepancy between the initial variational distribution and its improved version. We show that optimizing the VCD leads to better predicting performance on two latent variable
0
0
0
Thu Jun 07 2018
Machine Learning
Direct Optimization through for Discrete Variational Auto-Encoder
Reparametrization is an effective method for reducing the variance of their gradient estimates. We demonstrate the effectiveness of the direct loss minimization technique in variational autoencoders with both unstructured and structured discrete latent variables.
0
0
0
Sat Feb 06 2016
Machine Learning
R\'enyi Divergence Variational Inference
This paper introduces the variational R\'enyi bound (VR) that extends traditional variational inference to R\'eni's alpha-divergences. This new family of methods unifies a number of existing approaches. Experiments on Bayesian neural networks demonstrate the wide applicability of
0
0
0
Tue Nov 17 2020
Machine Learning
Recursive Inference for Variational Autoencoders
0
0
0