Controlling for discrete unmeasured confounding in nonlinear causal models

Patrick Burauel, Frederick Eberhardt, Michel Besserve
Proceedings of the Fourth Conference on Causal Learning and Reasoning, PMLR 275:991-1015, 2025.

Abstract

Unmeasured confounding is a major challenge for identifying causal relationships from non-experimental data. Here, we propose a method that can accommodate unmeasured discrete confounding. Extending recent identifiability results in deep latent variable models, we show theoretically that confounding can be detected and corrected under the assumption that the observed data is a piecewise affine transformation of a latent Gaussian mixture model and that the identity of the mixture components is confounded. We provide a flow-based algorithm to estimate this model and perform deconfounding. Experimental results on synthetic and real-world data provide support for the effectiveness of our approach.

Cite this Paper

BibTeX
@InProceedings{pmlr-v275-burauel25a, title = {Controlling for discrete unmeasured confounding in nonlinear causal models}, author = {Burauel, Patrick and Eberhardt, Frederick and Besserve, Michel}, booktitle = {Proceedings of the Fourth Conference on Causal Learning and Reasoning}, pages = {991--1015}, year = {2025}, editor = {Huang, Biwei and Drton, Mathias}, volume = {275}, series = {Proceedings of Machine Learning Research}, month = {07--09 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v275/main/assets/burauel25a/burauel25a.pdf}, url = {https://proceedings.mlr.press/v275/burauel25a.html}, abstract = {Unmeasured confounding is a major challenge for identifying causal relationships from non-experimental data. Here, we propose a method that can accommodate unmeasured discrete confounding. Extending recent identifiability results in deep latent variable models, we show theoretically that confounding can be detected and corrected under the assumption that the observed data is a piecewise affine transformation of a latent Gaussian mixture model and that the identity of the mixture components is confounded. We provide a flow-based algorithm to estimate this model and perform deconfounding. Experimental results on synthetic and real-world data provide support for the effectiveness of our approach.} }
Endnote
%0 Conference Paper %T Controlling for discrete unmeasured confounding in nonlinear causal models %A Patrick Burauel %A Frederick Eberhardt %A Michel Besserve %B Proceedings of the Fourth Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2025 %E Biwei Huang %E Mathias Drton %F pmlr-v275-burauel25a %I PMLR %P 991--1015 %U https://proceedings.mlr.press/v275/burauel25a.html %V 275 %X Unmeasured confounding is a major challenge for identifying causal relationships from non-experimental data. Here, we propose a method that can accommodate unmeasured discrete confounding. Extending recent identifiability results in deep latent variable models, we show theoretically that confounding can be detected and corrected under the assumption that the observed data is a piecewise affine transformation of a latent Gaussian mixture model and that the identity of the mixture components is confounded. We provide a flow-based algorithm to estimate this model and perform deconfounding. Experimental results on synthetic and real-world data provide support for the effectiveness of our approach.
APA
Burauel, P., Eberhardt, F. & Besserve, M.. (2025). Controlling for discrete unmeasured confounding in nonlinear causal models. Proceedings of the Fourth Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 275:991-1015 Available from https://proceedings.mlr.press/v275/burauel25a.html.

Related Material