Reproduce “Learning Summary Statistics for Bayesian Inference with Autoencoders” 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
Carlo Albert, Simone Ulzega, Firat Ozdemir, Fernando Perez-Cruz, Antonietta Mira
The authors develop learned summaries for likelihood-free calibration of stochastic models. The encoder combines parameter estimates with extra features, while the decoder receives simulation-noise information. In a 200-step, two-parameter test, three learned summaries yield a posterior closer to the Metropolis reference than two MLE summaries; INCA is somewhat wider than ENCA. This finding is limited to the tested models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.