ProximalFM: Amortized Proximal Causal Inference under Hidden Confounding

Key points

• Standard causal identification methods often assume no unmeasured confounding and can fail when relevant confounders are unobserved.1
• However, nonparametric proximal estimation can be challenging in practice: recovering causal estimands such as the conditional average treatment effect (CATE) requires solving an ill-posed integral equation that is data-hungry, hyperparameter-sensitive, and optimization-unstable.1
• Following the recent success of tabular foundation models in backdoor, instrumental variable, and frontdoor settings, we propose that prior-data fitted networks (PFNs) are uniquely suited to resolve this bottleneck.1
• Across a diverse suite of proximal regimes, ProximalFM achieves consistently strong CATE-estimation performance without dataset-specific tuning, with its largest advantage when latent confounding is substantial and the proxies are weakly informative; it also provides fast inference through a single amortized forward pass.1

Sources (1)

• [1]ProximalFM: Amortized Proximal Causal Inference under Hidden Confounding
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 10:07 AM

  Proximal causal inference instead uses proxy variables to identify effects under hidden confounding.

  Standard causal identification methods often assume no unmeasured confounding and can fail when relevant confounders are unobserved.

Extractive summary: sentences quoted from the sources.