Learning Optimal Interventions
This work addresses the challenge of safe and effective intervention design in domains like healthcare and education, though it is incremental as it builds on existing causal inference methods with tailored assumptions.
The paper tackles the problem of identifying beneficial interventions from observational data, proposing a conservative definition of optimal interventions to avoid harm, and develops efficient algorithms with theoretical guarantees, showing empirical improvements in gene perturbation and writing improvement applications.
Our goal is to identify beneficial interventions from observational data. We consider interventions that are narrowly focused (impacting few covariates) and may be tailored to each individual or globally enacted over a population. For applications where harmful intervention is drastically worse than proposing no change, we propose a conservative definition of the optimal intervention. Assuming the underlying relationship remains invariant under intervention, we develop efficient algorithms to identify the optimal intervention policy from limited data and provide theoretical guarantees for our approach in a Gaussian Process setting. Although our methods assume covariates can be precisely adjusted, they remain capable of improving outcomes in misspecified settings where interventions incur unintentional downstream effects. Empirically, our approach identifies good interventions in two practical applications: gene perturbation and writing improvement.