CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training

TL;DR

This paper introduces CausalGAN, an adversarial training framework for learning causal implicit generative models that can produce images consistent with causal graphs and handle both observational and interventional distributions.

Contribution

It proposes a novel two-stage training procedure and new architectures for causal generative modeling, enabling sampling from complex causal distributions.

Findings

WassersteinGAN effectively generates discrete labels.

CausalGAN can produce images conditioned on causal labels.

Models can simulate interventions beyond the training data.

Abstract

We propose an adversarial training procedure for learning a causal implicit generative model for a given causal graph. We show that adversarial training can be used to learn a generative model with true observational and interventional distributions if the generator architecture is consistent with the given causal graph. We consider the application of generating faces based on given binary labels where the dependency structure between the labels is preserved with a causal graph. This problem can be seen as learning a causal implicit generative model for the image and labels. We devise a two-stage procedure for this problem. First we train a causal implicit generative model over binary labels using a neural network consistent with a causal graph as the generator. We empirically show that WassersteinGAN can be used to output discrete labels. Later, we propose two new conditional GAN architectures, which we call CausalGAN and CausalBEGAN. We show that the optimal generator of the CausalGAN, given the labels, samples from the image distributions conditioned on these labels. The conditional GAN combined with a trained causal implicit generative model for the labels is then a causal implicit generative model over the labels and the generated image. We show that the proposed architectures can be used to sample from observational and interventional image distributions, even for interventions which do not naturally occur in the dataset.