Closed caxelrud closed 8 months ago
There are no published examples, but it is pretty straightforward to write a Bayesian neural network in the Infer.NET modelling language. A few people have tried this, but the inference results were not good. I think the reason is that the message-passing style inference in Infer.NET does not handle deterministic loops, and neural networks have lots of those.
Do you have or know any example related to Bayesian Neural Network (small, shallow cases) using Infer.net?