Adversarial Distribution Balancing for Counterfactual Reasoning
This addresses the problem of accurate counterfactual reasoning in fields like medicine, where treatment effects must be estimated from non-random data, representing a strong specific gain in causal inference.
The paper tackles the challenge of counterfactual reasoning in causal prediction, where outcomes are only observed for factual interventions, by proposing ADBCR to remove spurious causal relations using potential outcome estimates. It shows that ADBCR outperforms state-of-the-art methods on three benchmark datasets, with performance further improved by including unlabeled validation data.
The development of causal prediction models is challenged by the fact that the outcome is only observable for the applied (factual) intervention and not for its alternatives (the so-called counterfactuals); in medicine we only know patients' survival for the administered drug and not for other therapeutic options. Machine learning approaches for counterfactual reasoning have to deal with both unobserved outcomes and distributional differences due to non-random treatment administration. Unsupervised domain adaptation (UDA) addresses similar issues; one has to deal with unobserved outcomes -- the labels of the target domain -- and distributional differences between source and target domain. We propose Adversarial Distribution Balancing for Counterfactual Reasoning (ADBCR), which directly uses potential outcome estimates of the counterfactuals to remove spurious causal relations. We show that ADBCR outcompetes state-of-the-art methods on three benchmark datasets, and demonstrate that ADBCR's performance can be further improved if unlabeled validation data are included in the training procedure to better adapt the model to the validation domain.