Adversary-Free Counterfactual Prediction via Information-Regularized Representations

Shiqin Tang, Rong Feng, Shuxin Zhuang, Youzhi Zhang, Hongzong LI
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:5131-5139, 2026.

Abstract

We study counterfactual prediction under assignment bias and propose a mathematically grounded, information-theoretic approach that removes treatment–covariate dependence without adversarial training. Starting from a bound that links the counterfactual–factual risk gap to mutual information, we learn a stochastic representation $Z$ that is predictive of outcomes while minimizing $I(Z;T)$. We derive a tractable variational objective that upper-bounds the information term and couples it with a supervised decoder, yielding a stable, provably motivated training criterion. The framework extends naturally to dynamic settings by applying the information penalty to sequential representations at each decision time. We evaluate the method on controlled numerical simulations and a real-world clinical dataset, comparing against recent state-of-the-art balancing, reweighting, and adversarial baselines. Across metrics of likelihood, counterfactual error, and policy evaluation, our approach performs favorably while avoiding the training instabilities and tuning burden of adversarial schemes.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-tang26e, title = { Adversary-Free Counterfactual Prediction via Information-Regularized Representations }, author = {Tang, Shiqin and Feng, Rong and Zhuang, Shuxin and Zhang, Youzhi and LI, Hongzong}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {5131--5139}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/tang26e/tang26e.pdf}, url = {https://proceedings.mlr.press/v300/tang26e.html}, abstract = { We study counterfactual prediction under assignment bias and propose a mathematically grounded, information-theoretic approach that removes treatment–covariate dependence without adversarial training. Starting from a bound that links the counterfactual–factual risk gap to mutual information, we learn a stochastic representation $Z$ that is predictive of outcomes while minimizing $I(Z;T)$. We derive a tractable variational objective that upper-bounds the information term and couples it with a supervised decoder, yielding a stable, provably motivated training criterion. The framework extends naturally to dynamic settings by applying the information penalty to sequential representations at each decision time. We evaluate the method on controlled numerical simulations and a real-world clinical dataset, comparing against recent state-of-the-art balancing, reweighting, and adversarial baselines. Across metrics of likelihood, counterfactual error, and policy evaluation, our approach performs favorably while avoiding the training instabilities and tuning burden of adversarial schemes. } }
Endnote
%0 Conference Paper %T Adversary-Free Counterfactual Prediction via Information-Regularized Representations %A Shiqin Tang %A Rong Feng %A Shuxin Zhuang %A Youzhi Zhang %A Hongzong LI %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-tang26e %I PMLR %P 5131--5139 %U https://proceedings.mlr.press/v300/tang26e.html %V 300 %X We study counterfactual prediction under assignment bias and propose a mathematically grounded, information-theoretic approach that removes treatment–covariate dependence without adversarial training. Starting from a bound that links the counterfactual–factual risk gap to mutual information, we learn a stochastic representation $Z$ that is predictive of outcomes while minimizing $I(Z;T)$. We derive a tractable variational objective that upper-bounds the information term and couples it with a supervised decoder, yielding a stable, provably motivated training criterion. The framework extends naturally to dynamic settings by applying the information penalty to sequential representations at each decision time. We evaluate the method on controlled numerical simulations and a real-world clinical dataset, comparing against recent state-of-the-art balancing, reweighting, and adversarial baselines. Across metrics of likelihood, counterfactual error, and policy evaluation, our approach performs favorably while avoiding the training instabilities and tuning burden of adversarial schemes.
APA
Tang, S., Feng, R., Zhuang, S., Zhang, Y. & LI, H.. (2026). Adversary-Free Counterfactual Prediction via Information-Regularized Representations . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:5131-5139 Available from https://proceedings.mlr.press/v300/tang26e.html.

Related Material