Robust Entropy Search for Safe Efficient Bayesian Optimization
Abstract
The practical use of Bayesian Optimization (BO) in engineering applications imposes special requirements: high sampling efficiency on the one hand and finding a robust solution on the other hand. We address the case of adversarial robustness, where all parameters are controllable during the optimization process, but a subset of them is uncontrollable or even adversely perturbed at the time of application. To this end, we develop an efficient information-based acquisition function that we call Robust Entropy Search (RES). We empirically demonstrate its benefits in experiments on synthetic and real-life data. The results show that RES reliably finds robust optima, outperforming state-of-the-art algorithms.
Cite
Text
Weichert et al. "Robust Entropy Search for Safe Efficient Bayesian Optimization." Uncertainty in Artificial Intelligence, 2024.Markdown
[Weichert et al. "Robust Entropy Search for Safe Efficient Bayesian Optimization." Uncertainty in Artificial Intelligence, 2024.](https://mlanthology.org/uai/2024/weichert2024uai-robust/)BibTeX
@inproceedings{weichert2024uai-robust,
title = {{Robust Entropy Search for Safe Efficient Bayesian Optimization}},
author = {Weichert, Dorina and Kister, Alexander and Houben, Sebastian and Link, Patrick and Ernis, Gunar},
booktitle = {Uncertainty in Artificial Intelligence},
year = {2024},
pages = {3711-3729},
volume = {244},
url = {https://mlanthology.org/uai/2024/weichert2024uai-robust/}
}