DeepAI
Log In Sign Up

The Thermodynamic Variational Objective

06/28/2019
by   Vaden Masrani, et al.
2

We introduce the thermodynamic variational objective (TVO) for learning in both continuous and discrete deep generative models. The TVO arises from a key connection between variational inference and thermodynamic integration that results in a tighter lower bound to the log marginal likelihood than the standard variational evidence lower bound (ELBO), while remaining as broadly applicable. We provide a computationally efficient gradient estimator for the TVO that applies to continuous, discrete, and non-reparameterizable distributions and show that the objective functions used in variational inference, variational autoencoders, wake sleep, and inference compilation are all special cases of the TVO. We evaluate the TVO for learning of discrete and continuous variational auto encoders, and find it achieves state of the art for learning in discrete variable models, and outperform VAEs on continuous variable models without using the reparameterization trick.

READ FULL TEXT

page 1

page 2

page 3

page 4

11/12/2017

Alpha-Divergences in Variational Dropout

We investigate the use of alternative divergences to Kullback-Leibler (K...
06/18/2020

An adversarial algorithm for variational inference with a new role for acetylcholine

Sensory learning in the mammalian cortex has long been hypothesized to i...
06/23/2020

On the Relationship Between Active Inference and Control as Inference

Active Inference (AIF) is an emerging framework in the brain sciences wh...
04/26/2020

Notes on Icebreaker

Icebreaker [1] is new research from MSR that is able to achieve state of...
02/12/2018

Augment and Reduce: Stochastic Inference for Large Categorical Distributions

Categorical distributions are ubiquitous in machine learning, e.g., in c...
12/01/2017

Faithful Model Inversion Substantially Improves Auto-encoding Variational Inference

In learning deep generative models, the encoder for variational inferenc...
01/30/2023

A theory of continuous generative flow networks

Generative flow networks (GFlowNets) are amortized variational inference...