Bayesian Optimization Executable and Visualizable Application (BOXVIA) is a GUI-based application for Bayesian optimization. By using BOXVIA, users can perform Bayesian optimization and visualize functions obtained from the optimization process (i.e. mean function, its standard deviation, and acquisition function) without construction of a computing environment and programming skills. BOXVIA offers significant help for incorporating Bayesian optimization into your optimization problem.
Download an executable application
You can download executable files for BOXVIA from Releases.
Related paper
Akimitsu Ishii, Ryunosuke Kamijyo, Akinori Yamanaka, and Akiyasu Yamamoto, “BOXVIA: Bayesian optimization executable and visualizable application”, SoftwareX, Vol. 18 (2022/6), 101019. Link to ScienceDirect (Open Access Paper)
How to start BOXVIA
For using executable file
Extract the downloaded file and double-click on “BOXVIA” executable file.
The name of the executable is “BOXVIA.exe” on windows, “BOXVIA.app” on macOS, and “BOXVIA” on linux.
For running source code
Start with
python main.py
Dependencies
To start BOXVIA via the source codes, the following libraries are required.
If you use executable file, no dependencies are required.
- GPyOpt
- matplotlib
- pandas
- Dash
- Dash_bootstrap_components
You can install the dependencies by:
pip install -r requirements.txt
Note:
If you start BOXVIA via the source codes, the following two codes of
GPyOpt need to be modified.
- GPyOpt/core/bo.py
- GPyOpt/core/evaluators/batch_local_penalization.py
Please replace these codes with the codes contained in src/GPyOpt_modified in this repository.
Tutorial
Please see video tutorials uploaded on YouTube.
Video tutorial for 1D function
Video tutorial for 5D function
Further detailes are described in our paper.
Text-based tutorial has been posted on Qiita (in Japanese).
License
BSD License (3-clause BSD License)
Citation
@article{ISHII2022101019,
title = {BOXVIA: Bayesian optimization executable and visualizable application},
journal = {SoftwareX},
volume = {18},
pages = {101019},
year = {2022},
issn = {2352-7110},
doi = {https://doi.org/10.1016/j.softx.2022.101019},
url = {https://www.sciencedirect.com/science/article/pii/S2352711022000243},
author = {Akimitsu Ishii and Ryunosuke Kamijyo and Akinori Yamanaka and Akiyasu Yamamoto},
}