Differentiable Antithetic Sampling for Variance Reduction in Stochastic Variational Inference
Mike Wu, Noah Goodman, Stefano Ermon
Abstract
Stochastic optimization techniques are standard in variational inference algorithms. These methods estimate gradients by approximating expectations with independent Monte Carlo samples. In this paper, we explore a technique that uses correlated, but more representative, samples to reduce estimator variance. Specifically, we show how to generate antithetic samples that match sample moments with the true moments of an underlying importance distribution. Combining a differentiable antithetic sampler with modern stochastic variational inference, we showcase the effectiveness of this approach for learning a deep generative model. An implementation is available at https://github.com/mhw32/antithetic-vae-public.
BibTeX
@InProceedings{pmlr-v89-wu19c,
title = {Differentiable Antithetic Sampling for Variance Reduction in Stochastic Variational Inference},
author = {Wu, Mike and Goodman, Noah and Ermon, Stefano},
booktitle = {Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics},
pages = {2877--2886},
year = {2019},
editor = {Chaudhuri, Kamalika and Sugiyama, Masashi},
volume = {89},
series = {Proceedings of Machine Learning Research},
month = {16--18 Apr},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v89/wu19c/wu19c.pdf},
url = {https://proceedings.mlr.press/v89/wu19c.html},
abstract = {Stochastic optimization techniques are standard in variational inference algorithms. These methods estimate gradients by approximating expectations with independent Monte Carlo samples. In this paper, we explore a technique that uses correlated, but more representative, samples to reduce estimator variance. Specifically, we show how to generate antithetic samples that match sample moments with the true moments of an underlying importance distribution. Combining a differentiable antithetic sampler with modern stochastic variational inference, we showcase the effectiveness of this approach for learning a deep generative model. An implementation is available at https://github.com/mhw32/antithetic-vae-public.}
}