Causal Discovery from Interventional Data
摘要
In this Chapter we present an overview of how to discover the causal structure through interventions. We start by providing a formal definition of an intervention and describing the different types of interventions. Then we formalize the definition of an experiment and the assumptions that the search methods usually make. Finally, we describe fixed and adaptive search strategies to recover the causal structures via experiments, and provide worst case bounds on the required number of experiments for the different types of interventions.