Bayesian neural networks are a flexible, non-parametric method capable of modeling complex relationships. This software package provides a GPU-based framework for modeling and inference using Bayesian neural networks, capable of scaling to large datasets. For more information regarding the method, please see the following paper:
Bayesian Neural Networks for Genetic Association Studies of Complex Disease
The code is written in Python and built onto of Nvidia's CUDA GPU programming framework. The project code can be found at my github repository: