Double Control Variates for Gradient Estimation in Discrete Latent Variable Models

11/09/2021
by   Michalis K. Titsias, et al.
0

Stochastic gradient-based optimisation for discrete latent variable models is challenging due to the high variance of gradients. We introduce a variance reduction technique for score function estimators that makes use of double control variates. These control variates act on top of a main control variate, and try to further reduce the variance of the overall estimator. We develop a double control variate for the REINFORCE leave-one-out estimator using Taylor expansions. For training discrete latent variable models, such as variational autoencoders with binary latent variables, our approach adds no extra computational cost compared to standard training with the REINFORCE leave-one-out estimator. We apply our method to challenging high-dimensional toy examples and training variational autoencoders with binary latent variables. We show that our estimator can have lower variance compared to other state-of-the-art estimators.

READ FULL TEXT

page 1

page 2

page 3

page 4

02/19/2022

Gradient Estimation with Discrete Stein Operators

Gradient estimation – approximating the gradient of an expectation with ...
08/12/2022

Gradient Estimation for Binary Latent Variables via Gradient Variance Clipping

Gradient estimation is often necessary for fitting generative models wit...
08/12/2018

A Fourier View of REINFORCE

We show a connection between the Fourier spectrum of Boolean functions a...
04/01/2020

SUMO: Unbiased Estimation of Log Marginal Probability for Latent Variable Models

Standard variational lower bounds used to train latent variable models p...
10/07/2021

Bias-Variance Tradeoffs in Single-Sample Binary Gradient Estimators

Discrete and especially binary random variables occur in many machine le...
10/20/2020

VarGrad: A Low-Variance Gradient Estimator for Variational Inference

We analyse the properties of an unbiased gradient estimator of the ELBO ...
10/28/2021

Leveraging Recursive Gumbel-Max Trick for Approximate Inference in Combinatorial Spaces

Structured latent variables allow incorporating meaningful prior knowled...

Please sign up or login with your details

Forgot password? Click here to reset