berk-distributionalapproach-2020.pdf (10.89 MB)
A Distributional Approach towards Efficient and Versatile Bayesian Optimisation
thesis
posted on 2023-04-06, 05:34 authored by Julian Maxwell Andrew BerkThis study improved an algorithm for optimising costly black-box systems called Bayesian optimisation. It enhanced surrogate models Bayesian optimisation uses to select samples. It also introduced a method for altering the region from which the sample can be selected. These changes increased the algorithms efficiency, potentially reducing optimisation costs.
History
Pagination
91 p.Open access
- Yes