PhD Student, CISPA Saarbrücken
Compression-Driven Causal Discovery
Given multiple datasets over a fixed set of random variables, each collected from a different environment, we are interested in discovering the shared underlying causal network and the local interventions per environment, without assuming prior knowledge of which datasets are observational or interventional, and without assuming the shape of the causal dependencies. We formalize this problem using the Algorithmic Model of Causation, instantiate two consistent scores via the Minimum Description Length principle, and show under which conditions the network and interventions are identifiable. To efficiently discover causal networks and intervention targets in practice, we introduce the GLOBE and ORION algorithms, which we show outperform the state of the art in causal structure discovery over single as well as multiple environments.