Variational Causal Autoencoder for Interventional and Counterfactual Queries
We propose the Variational Causal Autoencoder (VCAUSE), a novel class of variational graph autoencoders for causal inference in the absence of hidden confounders, when only observational data and the causal graph are available. Without making any structural assumptions, VCAUSE mimics the necessary properties of a Structural Causal Model (SCM) to provide a framework for performing interventions (do-operator) and abduction-action-prediction steps. As a result, and as shown by our empirical results, VCAUSE provides a practical and accurate pipeline for estimating the interventional and counterfactual distributions of diverse SCMs. Finally, we apply VCAUSE to evaluate counterfactual fairness in classification problems and also to learn accurate and fair classifiers.
PDF Abstract