Skip to content

Multiobjective black-box optimization using gradient-boosted trees

License

Notifications You must be signed in to change notification settings

R-M-Lee/entmoot

 
 

Repository files navigation

Tests coverage Python versions Read the Docs

ENTMOOT (ENsemble Tree MOdel Optimization Tool) is a framework to perform Bayesian Optimization using tree-based surrogate models. Gradient-boosted tree models from lightgbm are combined with a distance-based uncertainty measure in a deterministic global optimization framework to optimize black-box functions. More details on the method here: https://arxiv.org/abs/2003.04774.

Documentation

The docs can be found here: https://entmoot.readthedocs.io/

How to reference ENTMOOT

When using any ENTMOOT for any publications please reference this software package as:

@article{thebelt2021entmoot,
  title={ENTMOOT: A framework for optimization over ensemble tree models},
  author={Thebelt, Alexander and Kronqvist, Jan and Mistry, Miten and Lee, Robert M and Sudermann-Merx, Nathan and Misener, Ruth},
  journal={Computers \& Chemical Engineering},
  volume={151},
  pages={107343},
  year={2021},
  publisher={Elsevier}
}

Authors

License

The ENTMOOT package is released under the BSD 3-Clause License. Please refer to the LICENSE file for details.

Acknowledgements

The support of BASF SE, Lugwigshafen am Rhein is gratefully acknowledged.

About

Multiobjective black-box optimization using gradient-boosted trees

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%