Work While They Sleep: Exploiting Evaluation Latency for Fully Bayesian Optimization
Abstract
Black-box optimization problems are ubiquitous across science and engineering, often dealing with expensive objective functions. This objective latency has two consequences during optimization: (i) the objective evaluation dominates execution time, and (ii) sample-efficient algorithms are crucial to accelerate development and avoid wasting resources. Bayesian optimization (BO) methods are the \textit{de facto} choice of planners for suggesting the next point to try. Standard BO fits the surrogate model's hyperparameters with a point estimate. Alternatively, a fully Bayesian approach uses model averaging to account for uncertainty over the hyperparameters, leading to better uncertainty estimates---useful in the low-data regime that is pervasive in BO. However, it is often prohibitively expensive and thus rarely used. In this work, we propose ELF-BO, an algorithm that uses the objective evaluation latency to headstart the computation of the next suggestion, allowing for fully Bayesian optimization without incurring substantial decision-time costs. This is done by sampling from the hyperparameter posterior \emph{while} the objective is being evaluated, only requiring reweighting of the samples once the objective value is observed. Across synthetic functions and real-world applications, we show that ELF-BO matches the performance of fully Bayesian methods while only incurring decision latency on par with or better than standard BO. Thus, ELF-BO makes fully Bayesian optimization practical in real-world use cases.
Related
- Regime-Conditioned Evaluation in Multi-Context Bayesian Optimization
- Exploiting Gradients in Bayesian Inference of Expensive Simulators
- Offline–Online Decoupled Architecture for Projected Microsecond-Latency and near Machine-Precision Caputo Fractional Derivative Evaluation, with an Edge-Computing Feasibility Study
- Transfer learning in bayesian optimization for aircraft design
- Pre-training of Bayesian Optimization Algorithm through Bayesian Optimization
BibTeX
@misc{sutter2026work,
title = {{Work While They Sleep: Exploiting Evaluation Latency for Fully Bayesian Optimization}},
author = {Gustavo Sutter and Alejandro Comas-Leon and David Holzmüller and Hao Wang and Luis Ricardez-Sandoval and Pascal Poupart and Agustinus Kristiadi},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.08969},
eprint = {2610.08969},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.08969},
}
RIS
TY - UNPB
TI - Work While They Sleep: Exploiting Evaluation Latency for Fully Bayesian Optimization
AU - Gustavo Sutter
AU - Alejandro Comas-Leon
AU - David Holzmüller
AU - Hao Wang
AU - Luis Ricardez-Sandoval
AU - Pascal Poupart
AU - Agustinus Kristiadi
PY - 2026
DA - 2026/10/08
PB - arXiv
DO - 10.48550/arxiv.2610.08969
UR - https://arxiv.org/abs/2610.08969
AB - Black-box optimization problems are ubiquitous across science and engineering, often dealing with expensive objective functions. This objective latency has two consequences during optimization: (i) the objective evaluation dominates execution time, and (ii) sample-efficient algorithms are crucial to accelerate development and avoid wasting resources. Bayesian optimization (BO) methods are the \textit{de facto} choice of planners for suggesting the next point to try. Standard BO fits the surrogate model's hyperparameters with a point estimate. Alternatively, a fully Bayesian approach uses model averaging to account for uncertainty over the hyperparameters, leading to better uncertainty estimates---useful in the low-data regime that is pervasive in BO. However, it is often prohibitively expensive and thus rarely used. In this work, we propose ELF-BO, an algorithm that uses the objective evaluation latency to headstart the computation of the next suggestion, allowing for fully Bayesian optimization without incurring substantial decision-time costs. This is done by sampling from the hyperparameter posterior \emph{while} the objective is being evaluated, only requiring reweighting of the samples once the objective value is observed. Across synthetic functions and real-world applications, we show that ELF-BO matches the performance of fully Bayesian methods while only incurring decision latency on par with or better than standard BO. Thus, ELF-BO makes fully Bayesian optimization practical in real-world use cases.
ER -