Reproduce “BayesIMP: Uncertainty Quantification for Causal Data Fusion” 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
Siu Lun Chau, Jean-François Ton, Javier González, Yee Whye Teh, Dino Sejdinovic
BAYESIMP estimates causal effects by combining treatment-to-mediator and mediator-to-outcome studies, carrying uncertainty from both through a kernel-embedding representation and a Gaussian-process response model. The authors evaluate it in synthetic experiments and a healthcare-inspired simulator, where it reaches the optimization target in fewer iterations than CBO. These results concern the reported tests; the method assumes known causal graphs and combines two graphs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.