BANNER

Bayesian Non-parametric Network Regression using variational Generalized Wishart Processes

This project consists of a gpflow 2 implementation of the variational Generalized Wishart Process , based on the Generalized Wishart Process. The implementation consists of the exact Wishart Process model and likelihood, as well as the factorized approximation. In addition, a multi output kernel is added which allows several input channels to share the same kernel (and thus learn the same lengthscale).

Contact

• Max Hinne
• Hester Huijsdens
• David Leeftink

Package requirements

package	version
gpflow	2.1.4
tensorflow	2.5.0
tensorflow_probability	0.12.1
tensorboard	2.5.0
matplotlib	3.3.2
numpy	1.19.2
h5py	3.1.0
scikit-learn	0.24.2
pandas	1.3.2
tqdm	4.62.3

Project structure

├── data                    # Folder for offline data
├── logs                    # Saving trained models and training logs     
├── analyses                # Training scripts and jupyter notebook examples
├── src                     # Source files
│   ├── models   
|   ├── kernels   
|   ├── likelihood   
└── README.md	

Tensorboard

Run "tensorboard --logdir logs/" in command line