No-Regret Bayesian Optimization with Stochastic Observation Failures
Abstract
We study Bayesian optimization problems where observation of the objective function fails stochastically, e.g., synthesis failures in materials development. For this problem, although several heuristic methods have been proposed, they do not have theoretical guarantees and sometimes deteriorate in practice. We propose two algorithms that have a trade-off relation between regret bounds and practical performance. The first one is the first no-regret algorithm for this problem. The second one shows superior practical performance; however, we need some modification of the algorithm to obtain a no-regret guarantee, which is slightly worse than the first one. We demonstrate the effectiveness of our methods in numerical experiments, including the simulation function motivated by quasi-crystal synthesis.
Cite
Text
Iwazaki et al. "No-Regret Bayesian Optimization with Stochastic Observation Failures." Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, 2025.Markdown
[Iwazaki et al. "No-Regret Bayesian Optimization with Stochastic Observation Failures." Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, 2025.](https://mlanthology.org/aistats/2025/iwazaki2025aistats-noregret/)BibTeX
@inproceedings{iwazaki2025aistats-noregret,
title = {{No-Regret Bayesian Optimization with Stochastic Observation Failures}},
author = {Iwazaki, Shogo and Tanabe, Tomohiko and Irie, Mitsuru and Takeno, Shion and Matsui, Kota and Inatsu, Yu},
booktitle = {Proceedings of The 28th International Conference on Artificial Intelligence and Statistics},
year = {2025},
pages = {415-423},
volume = {258},
url = {https://mlanthology.org/aistats/2025/iwazaki2025aistats-noregret/}
}