@inproceedings{hartford2017deep,
author = {Hartford, Jason and Lewis, Greg and Leyton-Brown, Kevin and Taddy, Matt},
title = {Deep IV: A Flexible Approach for Counterfactual Prediction},
booktitle = {Proceedings of Machine Learning Research},
year = {2017},
month = {August},
abstract = {Counterfactual prediction requires understanding causal relationships between so-called treatment and outcome variables. This paper provides a recipe for augmenting deep learning methods to accurately characterize such relationships in the presence of instrument variables (IVs) – sources of treatment randomization that are conditionally independent from the outcomes. Our IV specification resolves into two prediction tasks that can be solved with deep neural nets: a first-stage network for treatment prediction and a second-stage network whose loss function involves integration over the conditional treatment distribution. This Deep IV framework allows us to take advantage of off-the-shelf supervised learning techniques to estimate causal effects by adapting the loss function. Experiments show that it outperforms existing machine learning approaches.},
publisher = {PMLR},
url = {http://approjects.co.za/?big=en-us/research/publication/deep-iv-a-flexible-approach-for-counterfactual-prediction/},
volume = {70},
edition = {Proceedings of Machine Learning Research},
}