Reproduce “Estimating Causal Bounds from Observational Data With Neural Generative Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Max Zhu, Srivathsan Amruth, Sina Akbari, Qingyuan Zhao
The authors estimate treatment-effect ranges from observational data when hidden confounding leaves multiple causal models compatible with the observations. They train neural models at several nonzero penalties, then extrapolate estimated effects to the exact-fit limit. On synthetic benchmarks, the method improves on evaluated neural baselines, while analytical methods are exact on two instrumental-variable tasks; recovering bounds theoretically requires global minimizers.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.