analiza aiqbench
This research advances causal machine learning by proposing a ground-truth-free framework for selecting optimal heterogeneous treatment effect estimators without relying on unobservable true effects. By leveraging cross-fitted test statistics and sample splitting, the method ensures stable inference and significantly reduces false selections across standard benchmarks like ACIC and IHDP. The work addresses a critical reliability challenge in deploying causal AI models where validation data is inherently limited or unavailable.
Reliable Selection of Heterogeneous Treatment Effect Estimators
arXiv:2511.18464v2 Announce Type: replace-cross Abstract: We study the problem of selecting the best heterogeneous treatment effect (HTE) estimator from a collection of candidates in settings where the treatment effect is fundamentally unobserved. We cast estimator selection as a multiple testing problem and introduce a ground-truth-free procedure based on a cross-fitted, exponentially weighted test statistic. A key component of our method is a two-way sample splitting scheme that decouples nuisance estimation from weight learning and ensures the stability required for valid inference. Leveraging a stability-based central limit theorem, we establish asymptotic familywise error rate control under mild regularity conditions. Empirically, our procedure provides reliable error control while substantially reducing false selections compared with commonly used methods across ACIC 2016, IHDP, and Twins benchmarks, demonstrating that our method is feasible and powerful even without ground-truth treatment effects.
Abstract:We study the problem of selecting the best heterogeneous treatment effect (HTE) estimator from a collection of candidates in settings where the treatment effect is fundamentally unobserved. We cast estimator selection as a multiple testing problem and introduce a ground-truth-free procedure based on a cross-fitted, exponentially weighted test statistic. A key component of our method is a two-way sample splitting scheme that decouples nuisance estimation from weight learning and ensures the stability required for valid inference. Leveraging a stability-based central limit theorem, we establish asymptotic familywise error rate control under mild regularity conditions. Empirically, our procedure provides reliable error control while substantially reducing false selections compared with commonly used methods across ACIC 2016, IHDP, and Twins benchmarks, demonstrating that our method is feasible and powerful even without ground-truth treatment effects.
Machine Learning (stat.ML); Machine Learning (cs.LG)
arXiv:2511.18464 [stat.ML]
(or arXiv:2511.18464v2 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2511.18464 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 225 cuvinte
Subiecte conexe
Splitting the Difference: Interpretable Causal Forests for Treatment Effect Heterogeneity and Bias
arXiv · 16 sept. 2026
News Archive 2026
Quantum Computing Report · 25 iul. 2026
Barycentric Fused Gromov-Wasserstein Balancing for Causal Inference under Multiple Treatments
arXiv · 25 aug. 2026
Regression-Based Estimation of Causal Effects in the Presence of Selection Bias and Confounding
arXiv · 24 aug. 2026