Streaming sparse Gaussian process approximations

This repository contains an implementation of several online/streaming sparse GP approximations for regression and classification (Bui, Nguyen and Turner, NIPS 2017). In particular, osvgp.py implements the uncollapsed variational free-energy for regression and classification, and osgpr.py implements the collapsed variational free-energy and Power-EP energy for the regression case.

We also provide an implementation of the collapsed batch Power-EP sparse approximation of Bui, Yan and Turner (2017).

The code was tested using GPflow 2.5 and 2.6 and TensorFlow 2.5.

Usage

We provide several test scripts (regression and classification) to demonstrate the usage. Running these examples should the results similar to the following figures:

Contributors

Thang D. Bui
Cuong V. Nguyen
Richard E. Turner
ST John

References:

@inproceedings{BuiNguTur17,
  title =  {Streaming sparse {G}aussian process approximations},
  author =   {Bui, Thang D. and Nguyen, Cuong V. and Turner, Richard E.},
  booktitle = {Advances in Neural Information Processing Systems 30},
  year =   {2017}
}

@article{BuiYanTur16,
  title={A Unifying Framework for Sparse {G}aussian Process Approximation using {P}ower {E}xpectation {P}ropagation},
  author={Thang D. Bui and Josiah Yan and Richard E. Turner},
  journal={arXiv preprint arXiv:1605.07066},
  year={2016}
}