Active learning of causal probability trees

TL;DR

This paper introduces a Bayesian approach for learning probability trees that represent causal information, efficiently combining observational and interventional data to improve causal inference with limited interventions.

Contribution

It presents a novel Bayesian method for learning probability trees from mixed data, optimizing intervention choices based on expected information gain.

Findings

Efficient learning demonstrated on simulated data

Method outperforms baseline in limited intervention scenarios

Applicable to real-world causal inference tasks

Abstract

The past two decades have seen a growing interest in combining causal information, commonly represented using causal graphs, with machine learning models. Probability trees provide a simple yet powerful alternative representation of causal information. They enable both computation of intervention and counterfactuals, and are strictly more general, since they allow context-dependent causal dependencies. Here we present a Bayesian method for learning probability trees from a combination of interventional and observational data. The method quantifies the expected information gain from an intervention, and selects the interventions with the largest gain. We demonstrate the efficiency of the method on simulated and real data. An effective method for learning probability trees on a limited interventional budget will greatly expand their applicability.