Reproduce “Leveraging Self-Consistency for Data-Efficient Amortized Bayesian Inference” 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
Marvin Schmitt, Desi R. Ivanova, Daniel Habermann, Ullrich Köthe, Paul-Christian Bürkner, Stefan T. Radev
The authors introduce SC-NPLE, which trains neural posterior and likelihood approximations to agree on the data’s marginal likelihood across parameter values, adding a self-consistency loss to NPLE. On the Two Moons benchmark, SC-NPLE using 512 simulated training examples performs on par in posterior estimation with NPLE using 4,096. This is a benchmark result, not a general reduction in computation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.