Estimating Treatment Effects in Networks using Domain Adversarial Training

2025-10-24

AI Summary

In network causal inference, estimation of heterogeneous treatment effects (HTE) suffers from bias due to interference effects and unknown exposure mappings; additionally, the interaction between network homophily and treatment assignment mechanisms induces unobserved network-level covariate shift, further compromising estimator consistency. To address these challenges, we propose HINet—the first causal learning framework that explicitly models such network-level shift without requiring prior knowledge of exposure mappings. Methodologically, HINet integrates graph neural networks to capture higher-order network dependencies and employs domain-adversarial training to learn treatment-invariant representations across heterogeneous network structures, jointly mitigating interference and distribution shift. Evaluated on synthetic and semi-synthetic network datasets, HINet consistently outperforms state-of-the-art causal GNNs and conventional HTE estimators, demonstrating superior robustness and estimation accuracy under complex interference patterns.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Estimating heterogeneous treatment effects in network settings is complicated by interference, meaning that the outcome of an instance can be influenced by the treatment status of others. Existing causal machine learning approaches usually assume a known exposure mapping that summarizes how the outcome of a given instance is influenced by others' treatment, a simplification that is often unrealistic. Furthermore, the interaction between homophily -- the tendency of similar instances to connect -- and the treatment assignment mechanism can induce a network-level covariate shift that may lead to inaccurate treatment effect estimates, a phenomenon that has not yet been explicitly studied. To address these challenges, we propose HINet, a novel method that integrates graph neural networks with domain adversarial training. This combination allows estimating treatment effects under unknown exposure mappings while mitigating the impact of (network-level) covariate shift. An extensive empirical evaluation on synthetic and semi-synthetic network datasets demonstrates the effectiveness of our approach.

Problem

• Estimating treatment effects under unknown network interference exposure mappings
• Addressing network-level covariate shift from homophily-treatment interactions
• Mitigating interference effects in heterogeneous treatment effect estimation

Innovation

• Uses graph neural networks for network data
• Integrates domain adversarial training for covariate shift
• Estimates treatment effects under unknown exposure mappings