Title

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

Notification! We strongly recommend running BOXVIA via the executable file because the application has not yet support the latest versions of dependencies. If you run BOXVIA with its source code, carefully check the version of the dependencies.

Dependencies

To start BOXVIA via the source codes, the following libraries are required. If you use executable file, you do not need to install the dependencies yourself.

GPyOpt
matplotlib
pandas
Dash == 2.0.0
Dash_bootstrap_components == 1.0.0

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},
}

Developers' Affiliation

Yamanaka Research Group @ TUAT

BOXVIA

Install / Use

README