Outcome-adapted Automatic Debiased Machine Learning

2026-07-03

AI Summary

This work addresses the challenge of efficiently estimating linear functionals such as treatment effects in causal inference, where inadequate covariate representations often introduce bias. The authors propose AutoDML, an outcome-adaptive debiased machine learning method that, within a sample-splitting framework, learns sparse covariate representations predictive of the outcome but uninformative about the Riesz representer. Theoretically, this shared representation preserves outcome-relevant information while discarding components correlated with the Riesz representer, thereby achieving asymptotic efficiency. A neural network implementation integrates sparse representation learning into the AutoDML framework to optimize the structure of covariate embeddings. Empirical evaluations demonstrate that the proposed method significantly outperforms existing approaches on both synthetic data and the IHDP semi-synthetic benchmark, attaining state-of-the-art estimation accuracy.

Abstract

Parameters of interest in causal inference, such as treatment or policy effects, can often be expressed as linear functionals of an outcome regression function. Automatic debiased machine learning (AutoDML) is a unified framework for obtaining asymptotically normal estimators of such parameters, which requires estimation of both a regression function and a Riesz representer. Existing AutoDML neural network architectures, such as RieszNet and MADNet, use a shared intermediate covariate representation. However, it remains unclear whether this shared representation should be predictive of the Riesz representer or the outcome. We show that a shared representation of the covariates that preserves predictive power of the outcome while discarding information about the Riesz representer is asymptotically more efficient than the baseline AutoDML estimator that uses all covariates. Motivated by these results, we propose the outcome-adapted AutoDML estimator and establish its asymptotic behavior in a sample splitting framework. We provide a neural network implementation of the estimator that learns a sparse representation of the covariates that is predictive of the outcome but not predictive of the Riesz representer. We demonstrate the efficiency gains of our estimator over existing alternatives on synthetic data and achieve state-of-the-art estimation accuracy on the semi-synthetic IHDP benchmark dataset.

Problem

• causal inference
• automatic debiased machine learning
• Riesz representer
• outcome regression
• estimation efficiency

Innovation

• AutoDML
• Riesz representer
• outcome-adapted representation
• asymptotic efficiency
• neural network