Counterfactual Normalization: Proactively Addressing Dataset Shift Using Causal Mechanisms

Adarsh Subbaswamy, Suchi Saria
Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, PMLR R16:946-956, 2018.

Abstract

Predictive models can fail to generalize from training to deployment environments because of dataset shift, posing a threat to model re- liability in practice. As opposed to previous methods which use samples from the target distribution to reactively correct dataset shift, we propose using graphical knowledge of the causal mechanisms relating variables in a pre- diction problem to proactively remove variables that participate in spurious associations with the prediction target, allowing models to gen- eralize across datasets. To accomplish this, we augment the causal graph with latent counter- factual variables that account for the underlying causal mechanisms, and show how we can es- timate these variables. In our experiments we demonstrate that models using good estimates of the latent variables instead of the observed variables transfer better from training to tar- get domains with minimal accuracy loss in the training domain.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR16-subbaswamy18a, title = {Counterfactual Normalization: Proactively Addressing Dataset Shift Using Causal Mechanisms}, author = {Subbaswamy, Adarsh and Saria, Suchi}, booktitle = {Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence}, pages = {946--956}, year = {2018}, editor = {Globerson, Amir and Silva, Ricardo}, volume = {R16}, series = {Proceedings of Machine Learning Research}, month = {06--10 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r16/main/assets/subbaswamy18a/subbaswamy18a.pdf}, url = {https://proceedings.mlr.press/r16/subbaswamy18a.html}, abstract = {Predictive models can fail to generalize from training to deployment environments because of dataset shift, posing a threat to model re- liability in practice. As opposed to previous methods which use samples from the target distribution to reactively correct dataset shift, we propose using graphical knowledge of the causal mechanisms relating variables in a pre- diction problem to proactively remove variables that participate in spurious associations with the prediction target, allowing models to gen- eralize across datasets. To accomplish this, we augment the causal graph with latent counter- factual variables that account for the underlying causal mechanisms, and show how we can es- timate these variables. In our experiments we demonstrate that models using good estimates of the latent variables instead of the observed variables transfer better from training to tar- get domains with minimal accuracy loss in the training domain.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Counterfactual Normalization: Proactively Addressing Dataset Shift Using Causal Mechanisms %A Adarsh Subbaswamy %A Suchi Saria %B Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2018 %E Amir Globerson %E Ricardo Silva %F pmlr-vR16-subbaswamy18a %I PMLR %P 946--956 %U https://proceedings.mlr.press/r16/subbaswamy18a.html %V R16 %X Predictive models can fail to generalize from training to deployment environments because of dataset shift, posing a threat to model re- liability in practice. As opposed to previous methods which use samples from the target distribution to reactively correct dataset shift, we propose using graphical knowledge of the causal mechanisms relating variables in a pre- diction problem to proactively remove variables that participate in spurious associations with the prediction target, allowing models to gen- eralize across datasets. To accomplish this, we augment the causal graph with latent counter- factual variables that account for the underlying causal mechanisms, and show how we can es- timate these variables. In our experiments we demonstrate that models using good estimates of the latent variables instead of the observed variables transfer better from training to tar- get domains with minimal accuracy loss in the training domain. %Z Reissued by PMLR on 04 October 2026.
APA
Subbaswamy, A. & Saria, S.. (2018). Counterfactual Normalization: Proactively Addressing Dataset Shift Using Causal Mechanisms. Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R16:946-956 Available from https://proceedings.mlr.press/r16/subbaswamy18a.html. Reissued by PMLR on 04 October 2026.

Related Material