This code is tested using python 3.8
Installing the dependencies:
pip install -r requirements.txt
Example of a command to run the code:
python scripts/run.py --algo PDBO --problem zdt2 --n-var 4 --n-obj 2 --batch-size 2 4 8 16 --n-seed 25
The problem name, batch size, the number of variables, and the number of objectives can be changed. You can enter multiple batch sizes to evaluate in parallel. All results are saved in the '\result' folder.
The problem name is selected from the following options: zdt1, zdt2, zdt3, dtlz1, dtlz3, dtlz5, gtd
The HV results per iteration are computed during the optimization and are included in the results files.
In order to calculate the PFD, after obtaining the results use the following command:
python scripts/calculateParetoFrontDiversity.py
If you use this code please cite our paper:
@article{Ahmadianshalchi_Belakaria_Doppa_2024,
title={Pareto Front-Diverse Batch Multi-Objective Bayesian Optimization},
volume={38},
url={https://ojs.aaai.org/index.php/AAAI/article/view/28951},
DOI={10.1609/aaai.v38i10.28951},
number={10},
journal={Proceedings of the AAAI Conference on Artificial Intelligence},
author={Ahmadianshalchi, Alaleh and Belakaria, Syrine and Doppa, Janardhan Rao},
year={2024},
month={Mar.},
pages={10784-10794}
}