Coordinated Double Machine Learning

TL;DR

This paper introduces a coordinated deep neural network approach within double machine learning to improve treatment effect estimation bias, demonstrating enhanced empirical results on simulated and real datasets.

Contribution

It proposes a novel coordinated training algorithm for neural networks in double machine learning, reducing bias compared to traditional independent training methods.

Findings

Reduced estimation bias with coordinated neural network training.

Improved treatment effect estimates on simulated data.

Enhanced empirical performance on real-world datasets.

Abstract

Double machine learning is a statistical method for leveraging complex black-box models to construct approximately unbiased treatment effect estimates given observational data with high-dimensional covariates, under the assumption of a partially linear model. The idea is to first fit on a subset of the samples two non-linear predictive models, one for the continuous outcome of interest and one for the observed treatment, and then to estimate a linear coefficient for the treatment using the remaining samples through a simple orthogonalized regression. While this methodology is flexible and can accommodate arbitrary predictive models, typically trained independently of one another, this paper argues that a carefully coordinated learning algorithm for deep neural networks may reduce the estimation bias. The improved empirical performance of the proposed method is demonstrated through numerical experiments on both simulated and real data.