Approximation Based Variance Reduction for Reparameterization Gradients

Abstract

Flexible variational distributions improve variational inference but are harder to optimize. In this work we present a control variate that is applicable for any reparameterizable distribution with known mean and covariance, e.g. Gaussians with any covariance structure. The control variate is based on a quadratic approximation of the model, and its parameters are set using a double-descent scheme. We empirically show that this control variate leads to large improvements in gradient variance and optimization convergence for inference with non-factorized variational distributions.

Cite

Text

Geffner and Domke. "Approximation Based Variance Reduction for Reparameterization Gradients." Neural Information Processing Systems, 2020.

Markdown

[Geffner and Domke. "Approximation Based Variance Reduction for Reparameterization Gradients." Neural Information Processing Systems, 2020.](https://mlanthology.org/neurips/2020/geffner2020neurips-approximation/)

BibTeX

@inproceedings{geffner2020neurips-approximation,
  title     = {{Approximation Based Variance Reduction for Reparameterization Gradients}},
  author    = {Geffner, Tomas and Domke, Justin},
  booktitle = {Neural Information Processing Systems},
  year      = {2020},
  url       = {https://mlanthology.org/neurips/2020/geffner2020neurips-approximation/}
}