Curvature-aware Expected Free Energy as an Acquisition Function for Bayesian Optimization

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Meera, Ajith Anil, Kouw, Wouter
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908916968849408
author Meera, Ajith Anil
Kouw, Wouter
author_facet Meera, Ajith Anil
Kouw, Wouter
contents We propose an Expected Free Energy-based acquisition function for Bayesian optimization to solve the joint learning and optimization problem, i.e., optimize and learn the underlying function simultaneously. We show that, under specific assumptions, Expected Free Energy reduces to Upper Confidence Bound, Lower Confidence Bound, and Expected Information Gain. We prove that Expected Free Energy has unbiased convergence guarantees for concave functions. Using the results from these derivations, we introduce a curvature-aware update law for Expected Free Energy and show its proof of concept using a system identification problem on a Van der Pol oscillator. Through rigorous simulation experiments, we show that our adaptive Expected Free Energy-based acquisition function outperforms state-of-the-art acquisition functions with the least final simple regret and error in learning the Gaussian process.
format Preprint
id arxiv_https___arxiv_org_abs_2603_26339
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Curvature-aware Expected Free Energy as an Acquisition Function for Bayesian Optimization
Meera, Ajith Anil
Kouw, Wouter
Machine Learning
Robotics
Systems and Control
We propose an Expected Free Energy-based acquisition function for Bayesian optimization to solve the joint learning and optimization problem, i.e., optimize and learn the underlying function simultaneously. We show that, under specific assumptions, Expected Free Energy reduces to Upper Confidence Bound, Lower Confidence Bound, and Expected Information Gain. We prove that Expected Free Energy has unbiased convergence guarantees for concave functions. Using the results from these derivations, we introduce a curvature-aware update law for Expected Free Energy and show its proof of concept using a system identification problem on a Van der Pol oscillator. Through rigorous simulation experiments, we show that our adaptive Expected Free Energy-based acquisition function outperforms state-of-the-art acquisition functions with the least final simple regret and error in learning the Gaussian process.
title Curvature-aware Expected Free Energy as an Acquisition Function for Bayesian Optimization
topic Machine Learning
Robotics
Systems and Control
url https://arxiv.org/abs/2603.26339