Home > Research > Publications & Outputs > Inducing point allocation for sparse Gaussian p...

Research

Associated organisational units

Links

View graph of relations

Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Published

Standard

Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation. / Moss, Henry B; Ober, Sebastian W; Picheny, Victor.
International Conference on Artificial Intelligence and Statistics. Vol. 206 PMLR, 2023. p. 5213-5230 (Proceedings of Machine Learning Research).

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Conference contribution/Paper › peer-review

Harvard

Moss, HB, Ober, SW & Picheny, V 2023, Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation. in International Conference on Artificial Intelligence and Statistics. vol. 206, Proceedings of Machine Learning Research, PMLR, pp. 5213-5230, 26th International Conference on Artificial Intelligence and Statistics, AISTATS 2023, Valencia, Spain, 25/04/23. <https://proceedings.mlr.press/v206/moss23a.html>

APA

Moss, H. B., Ober, S. W., & Picheny, V. (2023). Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation. In International Conference on Artificial Intelligence and Statistics (Vol. 206, pp. 5213-5230). (Proceedings of Machine Learning Research). PMLR. https://proceedings.mlr.press/v206/moss23a.html

Vancouver

Moss HB, Ober SW, Picheny V. Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation. In International Conference on Artificial Intelligence and Statistics. Vol. 206. PMLR. 2023. p. 5213-5230. (Proceedings of Machine Learning Research).

Author

Moss, Henry B ; Ober, Sebastian W ; Picheny, Victor. / Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation. International Conference on Artificial Intelligence and Statistics. Vol. 206 PMLR, 2023. pp. 5213-5230 (Proceedings of Machine Learning Research).

Bibtex

@inproceedings{cae1d9ea8860405eb8b4235fa5eac100,
title = "Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation",
abstract = "Sparse Gaussian processes are a key component of high-throughput Bayesian optimisation (BO) loops; however, we show that existing methods for allocating their inducing points severely hamper optimisation performance. By exploiting the quality-diversity decomposition of determinantal point processes, we propose the first inducing point allocation strategy designed specifically for use in BO. Unlike existing methods which seek only to reduce global uncertainty in the objective function, our approach provides the local high-fidelity modelling of promising regions required for precise optimisation. More generally, we demonstrate that our proposed framework provides a flexible way to allocate modelling capacity in sparse models and so is suitable for a broad range of downstream sequential decision making tasks.",
author = "Moss, {Henry B} and Ober, {Sebastian W} and Victor Picheny",
year = "2023",
month = apr,
day = "25",
language = "English",
volume = "206",
series = "Proceedings of Machine Learning Research",
publisher = "PMLR",
pages = "5213--5230",
booktitle = "International Conference on Artificial Intelligence and Statistics",
note = "26th International Conference on Artificial Intelligence and Statistics, AISTATS 2023 ; Conference date: 25-04-2023 Through 27-04-2023",

}

RIS

TY - GEN

T1 - Inducing point allocation for sparse Gaussian processes in high-throughput Bayesian optimisation

AU - Moss, Henry B

AU - Ober, Sebastian W

AU - Picheny, Victor

PY - 2023/4/25

Y1 - 2023/4/25

N2 - Sparse Gaussian processes are a key component of high-throughput Bayesian optimisation (BO) loops; however, we show that existing methods for allocating their inducing points severely hamper optimisation performance. By exploiting the quality-diversity decomposition of determinantal point processes, we propose the first inducing point allocation strategy designed specifically for use in BO. Unlike existing methods which seek only to reduce global uncertainty in the objective function, our approach provides the local high-fidelity modelling of promising regions required for precise optimisation. More generally, we demonstrate that our proposed framework provides a flexible way to allocate modelling capacity in sparse models and so is suitable for a broad range of downstream sequential decision making tasks.

AB - Sparse Gaussian processes are a key component of high-throughput Bayesian optimisation (BO) loops; however, we show that existing methods for allocating their inducing points severely hamper optimisation performance. By exploiting the quality-diversity decomposition of determinantal point processes, we propose the first inducing point allocation strategy designed specifically for use in BO. Unlike existing methods which seek only to reduce global uncertainty in the objective function, our approach provides the local high-fidelity modelling of promising regions required for precise optimisation. More generally, we demonstrate that our proposed framework provides a flexible way to allocate modelling capacity in sparse models and so is suitable for a broad range of downstream sequential decision making tasks.

M3 - Conference contribution/Paper

VL - 206

T3 - Proceedings of Machine Learning Research

SP - 5213

EP - 5230

BT - International Conference on Artificial Intelligence and Statistics

PB - PMLR

T2 - 26th International Conference on Artificial Intelligence and Statistics, AISTATS 2023

Y2 - 25 April 2023 through 27 April 2023

ER -