Neural Kernel Network

This code is jointly contributed by Shengyang Sun, Guodong Zhang, Chaoqi Wang and Wenyuan Zeng

Introduction

Code for "Differentiable Compositional Kernel Learning for Gaussian Processes" (https://arxiv.org/abs/1806.04326)

Dependencies

This project runs with Python 3.6. Before running the code, you have to install

• Tensorflow
• GPflow-Slim

Experiments

Below we shows some examples to run the experiments. We also provide experiment figures and logging files in results folder, as a reference.

Time Series

python exp/time-series.py --name airline --kern nkn

Regression

python exp/regression.py --data energy --split uci_woval --kern nkn
python exp/regression.py --data energy --split uci_woval_pca --kern nkn

Bayesian Optimization

python exp/bayes-opt.py --name sty --kern nkn --run 0

Texture Extrapolation

python exp/texture.py --data pave --kern nkn

Citation

To cite this work, please use

@article{sun2018differentiable,
  title={Differentiable Compositional Kernel Learning for Gaussian Processes},
  author={Sun, Shengyang and Zhang, Guodong and Wang, Chaoqi and Zeng, Wenyuan and Li, Jiaman and Grosse, Roger},
  journal={arXiv preprint arXiv:1806.04326},
  year={2018}
}