Deconfounding via Profiled Transfer Learning

2025-08-15

AI Summary

Unobserved confounding is a primary source of bias in regression-based causal effect estimation. This paper proposes ProTrans, a transfer learning framework leveraging profiled residuals to mitigate latent confounding in the target dataset by exploiting source datasets with similar confounding structures. Its key contributions are: (1) modeling shared confounding patterns between source and target domains via profiled residuals—without assuming knowledge of the true confounding structure; (2) designing a robust source selection mechanism to effectively filter out uninformative or harmful sources; and (3) enabling unbiased treatment effect estimation without requiring instrumental or proxy variables. We establish theoretical guarantees showing that ProTrans achieves the minimax optimal convergence rate under mild regularity conditions. Extensive simulations and empirical analyses demonstrate its effectiveness in bias reduction, estimation accuracy, and robustness to heterogeneous or noisy source data.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Responsible Web: Algorithmic accountability and transparency on the web

Abstract

Unmeasured confounders are a major source of bias in regression-based effect estimation and causal inference. In this paper, we advocate a new profiled transfer learning framework, ProTrans, to address confounding effects in the target dataset, when additional source datasets that possess similar confounding structures are available. We introduce the concept of profiled residuals to characterize the shared confounding patterns between source and target datasets. By incorporating these profiled residuals into the target debiasing step, we effectively mitigates the latent confounding effects. We also propose a source selection strategy to enhance robustness of ProTrans against noninformative sources. As a byproduct, ProTrans can also be utilized to estimate treatment effects when potential confounders exist, without the use of auxiliary features such as instrumental or proxy variables, which are often challenging to select in practice. Theoretically, we prove that the resulting estimated model shift from sources to target is confounding-free without any assumptions imposed on the true confounding structure, and that the target parameter estimation achieves the minimax optimal rate under mild conditions. Simulated and real-world experiments validate the effectiveness of ProTrans and support the theoretical findings.

Problem

• Addresses bias from unmeasured confounders in effect estimation
• Proposes transfer learning to mitigate latent confounding effects
• Estimates treatment effects without instrumental or proxy variables

Innovation

• Profiled transfer learning for confounding mitigation
• Profiled residuals characterize shared confounding patterns
• Source selection enhances robustness against noninformative data