Reproduce “Structural Causal Bottleneck 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

Simon Bing, Jonas Wahl, Jakob Runge

Structural Causal Bottleneck Models (SCBMs) estimate causal effects by replacing high-dimensional variables with low-dimensional summaries tailored to individual causal mechanisms. In synthetic linear adjustment experiments, conditioning on a 5-dimensional outcome-side bottleneck instead of a 100-dimensional mediator or confounder lowers finite-sample MSE. These results use oracle bottlenecks; the theory assumes a known graph and essentially exact low-rank mechanisms.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.