Latent Variable Models for Bayesian Causal Discovery

TL;DR

This paper introduces a novel latent variable decoder model called Decoder BCD for Bayesian causal discovery, demonstrating its effectiveness in synthetic data experiments for recovering causal structures.

Contribution

It proposes a new model for learning causal representations from high-dimensional data and explores its application in both supervised and unsupervised settings.

Findings

Using known intervention targets improves causal inference accuracy.

The model effectively recovers causal structures in synthetic experiments.

Intervention labels aid in unsupervised Bayesian causal discovery.

Abstract

Learning predictors that do not rely on spurious correlations involves building causal representations. However, learning such a representation is very challenging. We, therefore, formulate the problem of learning a causal representation from high dimensional data and study causal recovery with synthetic data. This work introduces a latent variable decoder model, Decoder BCD, for Bayesian causal discovery and performs experiments in mildly supervised and unsupervised settings. We present a series of synthetic experiments to characterize important factors for causal discovery and show that using known intervention targets as labels helps in unsupervised Bayesian inference over structure and parameters of linear Gaussian additive noise latent structural causal models.