DeepMatch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training

Nathan Kallus
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:5067-5077, 2020.

Abstract

We study optimal covariate balance for causal inferences from observational data when rich covariates and complex relationships necessitate flexible modeling with neural networks. Standard approaches such as propensity weighting and matching/balancing fail in such settings due to miscalibrated propensity nets and inappropriate covariate representations, respectively. We propose a new method based on adversarial training of a weighting and a discriminator network that effectively addresses this methodological gap. This is demonstrated through new theoretical characterizations and empirical results on both synthetic and clinical data showing how causal analyses can be salvaged in such challenging settings.

Cite this Paper

BibTeX
@InProceedings{pmlr-v119-kallus20a, title = {{D}eep{M}atch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training}, author = {Kallus, Nathan}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {5067--5077}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/kallus20a/kallus20a.pdf}, url = {https://proceedings.mlr.press/v119/kallus20a.html}, abstract = {We study optimal covariate balance for causal inferences from observational data when rich covariates and complex relationships necessitate flexible modeling with neural networks. Standard approaches such as propensity weighting and matching/balancing fail in such settings due to miscalibrated propensity nets and inappropriate covariate representations, respectively. We propose a new method based on adversarial training of a weighting and a discriminator network that effectively addresses this methodological gap. This is demonstrated through new theoretical characterizations and empirical results on both synthetic and clinical data showing how causal analyses can be salvaged in such challenging settings.} }
Endnote
%0 Conference Paper %T DeepMatch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training %A Nathan Kallus %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-kallus20a %I PMLR %P 5067--5077 %U https://proceedings.mlr.press/v119/kallus20a.html %V 119 %X We study optimal covariate balance for causal inferences from observational data when rich covariates and complex relationships necessitate flexible modeling with neural networks. Standard approaches such as propensity weighting and matching/balancing fail in such settings due to miscalibrated propensity nets and inappropriate covariate representations, respectively. We propose a new method based on adversarial training of a weighting and a discriminator network that effectively addresses this methodological gap. This is demonstrated through new theoretical characterizations and empirical results on both synthetic and clinical data showing how causal analyses can be salvaged in such challenging settings.
APA
Kallus, N.. (2020). DeepMatch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:5067-5077 Available from https://proceedings.mlr.press/v119/kallus20a.html.

Related Material