In this note we explore unsupervised deep-learning algorithms for simulating non-linear structural equation models from observational training data. The algorithms described here enable automated sampling of latent space distributions, and are capable of generating interventional conditional probabilities that are often faithful to the ground truth distributions well beyond the range of data contained in the training set. These methods could in principle be used in conjunction with any existing autoencoder that produces a latent space representation containing causal graph structures.
View on arXiv