Deep Gaussian Processes

The authors propose a new method for Bayesian causal inference, combining order-based MCMC structure learning with gradient-based graph learning. The problem of inferring causal structure is decomposed into inferring a…