Bayes Neural Network

Implementation of Bayes by Back-propagate.

Basic usage

• Building an inference net

  from bayes_network import BayesMLP
  device = 'cuda' 
  net = BayesMLP(device = device)
  net.load_state_dict(PATH_TO_WEIGHT)
  net.forward(image, sample = False) 

Note that the model BayesMLP is designed for the mnist classification.