Bayesian Optimization: Theory and Practice Using Python

© The Author(s), under exclusive license to APress Media, LLC, part of Springer Nature 2023

P. LiuBayesian Optimizationhttps://doi.org/10.1007/978-1-4842-9063-7_3

3. Bayesian Decision Theory and Expected Improvement

Peng Liu¹  

(1)

Singapore, Singapore

 

The previous chapter used Gaussian processes (GP) as the surrogate model to approximate the underlying objective function. GP is a flexible framework that provides uncertainty estimates in the form of probability distributions over plausible functions across the entire domain. We could then resort to the closed-form posterior predictive distributions at proposed locations to obtain an educated guess on the potential observations.

However, it is not the only choice of surrogate model used in Bayesian ...