[NeurIPS 2023] Bayesian Optimisation of Functions on Graphs

Code repository for NeurIPS 2023 paper Bayesian Optimisation of Functions on Graphs.

If you find the paper or the codebase useful to your research, please cite:

@inproceedings{
wan2023bayesian,
title={Bayesian Optimisation of Functions on Graphs},
author={Xingchen Wan and Pierre Osselin and Henry Kenlay and Binxin Ru and Michael A Osborne and Xiaowen Dong},
booktitle={Thirty-seventh Conference on Neural Information Processing Systems},
year={2023},
url={https://openreview.net/forum?id=UuNd9A6noD}
}

Create virtual env & install dependencies

conda create -n graph
conda install networkx numpy pandas matplotlib seaborn jupyterlab
conda install botorch -c pytorch -c gpytorch -c conda-forge
conda activate graph
pip install ndlib

Run

Use the following code block to run a single trial:

from search.run_one_replicate import run_one_replication

save_dir = "./logs/synthetic/"
seed = 0
problem_name = "small_ba_betweenness"       # defines the problem
label = "ei_ego_network_polynomial"                  # defines the method

run_one_replication(
    label,
    seed=seed,
    problem_name=problem_name,
    save_path=save_dir,
    batch_size=1,
    n_initial_points=10,
)

To run through command line, define a configuration in the folder config/config.yaml and run the command:

python main.py --config config

For example, to run the centrality experiment with BA graphs, run:

python main.py --config centrality_ba