andreaponti5 / bows

Code for the paper: Wasserstein enabled Bayesian optimization of composite functions.

Bayesian Optimization in Wasserstein Space

This repository contains the code of the algorithm BOWS used in the following paper:

Candelieri, A., Ponti, A. & Archetti, F. Wasserstein enabled Bayesian optimization of composite functions. J Ambient Intell Human Comput (2023). https://doi.org/10.3390/math11102342

Python dependencies

Use the requirements.txt file as reference.
You can automatically install all the dependencies using the following command.

pip install -r requirements.txt

How to use the code

There are two main entrypoints:

• run_bo.py: run the experiments using the standard BO algorithm.
• run_bows.py: run the experiments using the BOWS algorithm.

In both scripts, it is possible to modify the test function as well as the number of variables.

How to cite us

If you use this repository, please cite the following paper:

Candelieri, A., Ponti, A. & Archetti, F. Wasserstein enabled Bayesian optimization of composite functions. J Ambient Intell Human Comput (2023). https://doi.org/10.3390/math11102342

@Article{Candelieri2023,
  AUTHOR = {Candelieri, Antonio and Ponti, Andrea and Archetti, Francesco},
  TITLE = {Wasserstein enabled Bayesian optimization of composite functions},
  JOURNAL = {Journal of Ambient Intelligence and Humanized Computing},
  YEAR = {2023},
  URL = {https://doi.org/10.1007/s12652-023-04640-7},
  ISSN = {1868-5145},
  DOI = {10.1007/s12652-023-04640-7}
}