Bayesian Graph Convolution
Bayesian Graph Convolution implementation with Pytorch Geometric
As part of My Project @APPCAIR, I worked on Bayesian Deep learning and implemented a Bayesian Graph convolution Network with Pytorch Geometric. The weights of the Graph convolution are from a gaussian distribution. Most of the difficult part is done by Pytorch Geometric (kudos to the library maintainers and contributors).
TODO : Let users choose the weight distribution. Currently only gaussian distribution is implemented.