Interpretable Causal Discovery via Causal-Effect Constraints

Title: Interpretable Causal Discovery via Causal-Effect Constraints Source: arxiv Bayesian causal discovery is framed as a conditional inference problem to incorporate causal-effect constraints, such as specific magnitude requirements for edges. The method employs rare-event estimation to drive particle populations toward constrained regions of the joint graph-parameter space, overcoming computational failures in standard approaches when constraints have low posterior mass.