Reproduce “Shrinkage prior for implicit 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

Shreya Sinha-Roy, Christian P. Robert, Ritabrata Dutta

The authors develop a likelihood-free Bayesian method for neural simulators, combining an energy-score posterior with a global-local horseshoe prior to learn network weights from scarce observations. They update weights approximately and sample shrinkage variables conditionally. In Hopper dynamics prediction, the method outperforms Gaussian and Student-t prior alternatives in reported data-scarce tests; the study does not evaluate downstream policy performance.

Summary by alphaXiv.

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