Cycle-Consistent Adversarial Learning as Approximate Bayesian Inference

TL;DR

This paper presents a Bayesian inference framework for learning interdomain correspondences without paired data, unifying cycle-consistent adversarial learning with variational inference in implicit latent variable models.

Contribution

It introduces a flexible implicit latent variable model and a novel VI algorithm based on symmetric KL divergence, linking CYCLEGAN to Bayesian inference.

Findings

CYCLEGAN models are derived as a special case of the proposed VI framework.

The new VI algorithm effectively models unpaired data correspondence.

The framework offers a probabilistic interpretation of cycle-consistent adversarial learning.

Abstract

We formalize the problem of learning interdomain correspondences in the absence of paired data as Bayesian inference in a latent variable model (LVM), where one seeks the underlying hidden representations of entities from one domain as entities from the other domain. First, we introduce implicit latent variable models, where the prior over hidden representations can be specified flexibly as an implicit distribution. Next, we develop a new variational inference (VI) algorithm for this model based on minimization of the symmetric Kullback-Leibler (KL) divergence between a variational joint and the exact joint distribution. Lastly, we demonstrate that the state-of-the-art cycle-consistent adversarial learning (CYCLEGAN) models can be derived as a special case within our proposed VI framework, thus establishing its connection to approximate Bayesian inference methods.