batch-bkb is the first Bayesian optimization (a.k.a. Gaussian process or bandit optimization) algorithm that is both provably no-regret and guaranteed to run in near-linear time time.
This repository contains an implementation of the algorithm as described in the ICML 2020 paper "Near-linear time Gaussian process optimization with adaptive batching and resparsification" by Calandriello Daniele, Luigi Carratino, Alessandro Lazaric, Michal Valko and Lorenzo Rosasco,
This repository also contains an implementation of bkb as described in the COLT 2019 paper
"Gaussian Process Optimization with Adaptive Sketching: Scalable and No Regret"
by Calandriello Daniele,
Luigi Carratino, Alessandro Lazaric, Michal Valko and Lorenzo Rosasco
| Link | Resource |
|---|---|
| ArXiv | Paper |
| Poster | Poster |
Our code requires access to a scientific python stack, including numpy, scipy, and sklearn
An example usage can be find in the file example_batch_bkb.py and example_bkb.py
If you use batch-bkb or the related experiments code please cite:
@incollection{icml2020bbkb,
title = {Near-linear time Gaussian process optimization with adaptive batching and resparsification},
author = {Daniele Calandriello and Luigi Carratino and Alessandro Lazaric and Michal Valko and Lorenzo Rosasco},
booktitle = {Proceedings of the 37th International Conference on Machine Learning},
year = {2020},
}
@InProceedings{pmlr-v99-calandriello19a,
title = {Gaussian Process Optimization with Adaptive Sketching: Scalable and No Regret},
author = {Calandriello, Daniele and Carratino, Luigi and Lazaric, Alessandro and Valko, Michal and Rosasco, Lorenzo},
booktitle = {Proceedings of the Thirty-Second Conference on Learning Theory},
pages = {533--557},
year = {2019},
editor = {Beygelzimer, Alina and Hsu, Daniel},
volume = {99},
series = {Proceedings of Machine Learning Research},
month = {25--28 Jun},
publisher = {PMLR},
}
For any question, you can contact daniele.calandriello@iit.it or luigi.carratino@dibris.unige.it