Bayesian Convolutional Neural Network based on Bayes by Backprop in PyTorch
Switch branches/tags
Nothing to show
Clone or download

README.md

Bayesian CNN

Implementation of Bayes by Backprop in a convolutional neural network.

One convolutional layer with distributions over weights in each filter

Distribution over weights in a CNN's filter.

Fully Bayesian perspective of an entire CNN

Distributions must be over weights in convolutional layers and weights in fully-connected layers.

Results

Results on MNIST, CIFAR-10 and CIFAR-100 with 3Conv3FC

Results MNIST, CIFAR-10 and CIFAR-100 with 3Conv3FC

Please cite:

@ARTICLE{2018arXiv180605978S,
   author = {{Shridhar}, K. and {Laumann}, F. and {Llopart Maurin}, A. and 
	{Liwicki}, M.},
    title = "{Bayesian Convolutional Neural Networks}",
  journal = {ArXiv e-prints},
archivePrefix = "arXiv",
   eprint = {1806.05978},
 keywords = {Computer Science - Machine Learning, Computer Science - Computer Vision and Pattern Recognition, Computer Science - Neural and Evolutionary Computing, Statistics - Machine Learning},
     year = 2018,
    month = jun,
   adsurl = {http://adsabs.harvard.edu/abs/2018arXiv180605978S},
  adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}