A Bayesian Inference over Neural Networks On a supervised model parameterized by W, we seek to infer the conditional distribution W | D