AI Chat AI Image Generator AI Video Text to Speech

Variational Determinant Estimation with Spherical Normalizing Flows

12/24/2020

∙

by Simon Passenheim, et al.

∙

∙

This paper introduces the Variational Determinant Estimator (VDE), a variational extension of the recently proposed determinant estimator discovered by arXiv:2005.06553v2. Our estimator significantly reduces the variance even for low sample sizes by combining (importance-weighted) variational inference and a family of normalizing flows which allow density estimation on hyperspheres. In the ideal case of a tight variational bound, the VDE becomes a zero variance estimator, and a single sample is sufficient for an exact (log) determinant estimate.

Simon Passenheim
1 publication
Emiel Hoogeboom
19 publications

research

∙ 03/27/2017

Sticking the Landing: Simple, Lower-Variance Gradient Estimators for Variational Inference

We propose a simple and general variant of the standard reparameterized ...

0 Geoffrey Roeder, et al. ∙

research

∙ 06/18/2012

Tighter Variational Representations of f-Divergences via Restriction to Probability Measures

We show that the variational representations for f-divergences currently...

0 Avraham Ruderman, et al. ∙

research

∙ 10/20/2020

VarGrad: A Low-Variance Gradient Estimator for Variational Inference

We analyse the properties of an unbiased gradient estimator of the ELBO ...

0 Lorenz Richter, et al. ∙

research

∙ 04/04/2022

Discretely Indexed Flows

In this paper we propose Discretely Indexed flows (DIF) as a new tool fo...

0 Elouan Argouarc'h, et al. ∙

research

∙ 06/17/2022

Path-Gradient Estimators for Continuous Normalizing Flows

Recent work has established a path-gradient estimator for simple variati...

0 Lorenz Vaitl, et al. ∙

research

∙ 02/27/2023

U-Statistics for Importance-Weighted Variational Inference

We propose the use of U-statistics to reduce variance for gradient estim...

0 Javier Burroni, et al. ∙

research

∙ 09/27/2018

Variance reduction properties of the reparameterization trick

The reparameterization trick is widely used in variational inference as ...

0 Ming Xu, et al. ∙