Reproduce “Semi-supervised Invertible Neural Operators for Bayesian Inverse Problems” 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
Sebastian Kaltenbach, Paris Perdikaris, Phaedon-Stelios Koutsourelakis
The authors introduce an invertible DeepONet that uses a RealNVP branch to support forward PDE prediction and map solution observations back toward inputs. They estimate posterior distributions by approximating transformed prior features with a finite Gaussian mixture. In a 100-dimensional reaction–diffusion example, posterior means and uncertainty bands visually agree with NUTS MCMC in two randomly chosen cases only.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.