BIhNNs: Bayesian Inference with Hamiltonian Neural Networks

BIhNNS is a solution for the problem of Bayesian inference being computationally expensive. It accelerates Bayesian inference by using Hamiltonian Neural Networks (HNNs) within the widely adopted sampling methods, Hamiltonian Monte Carlo (HMC) and the No-U-Turn-Sampler (NUTS). These methods typically require costly gradient estimations, but using HNNs within them does not, thus making the process more efficient. The research and code are intended to be used in various fields, such as engineering, computational science, computer science, and machine learning. 

GitHub repository: https://github.com/IdahoLabResearch/BIhNNs