Wasserstein Variational Inference

Abstract

This paper introduces Wasserstein variational inference, a new form of approximate Bayesian inference based on optimal transport theory. Wasserstein variational inference uses a new family of divergences that includes both f-divergences and the Wasserstein distance as special cases. The gradients of the Wasserstein variational loss are obtained by backpropagating through the Sinkhorn iterations. This technique results in a very stable likelihood-free training method that can be used with implicit distributions and probabilistic programs. Using the Wasserstein variational inference framework, we introduce several new forms of autoencoders and test their robustness and performance against existing variational autoencoding techniques.

arXiv comment8 pages, 1 figure

References in corpus (10)

• Stochastic Backpropagation and Approximate Inference in Deep Generative Models
• Improved Training of Wasserstein GANs
• Adversarially Learned Inference
• Variational Autoencoder for Deep Learning of Images, Labels and Captions
• Edward: A library for probabilistic modeling, inference, and criticism
• Hierarchical Implicit Models and Likelihood-Free Variational Inference
• Variational Inference using Implicit Distributions
• Rényi Divergence Variational Inference
• Operator Variational Inference
• Advances in Variational Inference