EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1924. Bayesian optimisation for facility location problems

Invited abstract in session WD-28: Advancements of OR-analytics in statistics, machine learning and data science 11, stream Advancements of OR-analytics in statistics, machine learning and data science.

Wednesday, 14:30-16:00
Room: 065 (building: 208)

Authors (first author is the speaker)

1. Niyati Seth
Mathematics and Statistics, University College Dublin
2. Michael Fop
School of Mathematics and Statistics, University College Dublin

Abstract

We study the facility location problem under uncertainty where the decisions on where to locate facilities often require estimating different uncertain parameters associated with the problem, for example costs, supply, demand, distances, etc. which may fluctuate with time and circumstances. The recent body of literature has been towards incorporating this uncertainty at various levels.
Following and extending this branch of literature, our research aims at developing facility location procedures which account for uncertainty in the model and in the data. To do so, we extend Bayesian optimisation methods for combinatorial structures(BOCS) to solve the facility location problems.
Our research presents a novel approach by extending the BOCS method to account for constraints, uncertainty, grid structure and interaction of the allocation nodes. Incorporating constraints in combinatorial domain within the framework of BOCS is challenging. Hence, we propose an extension to the framework by introducing probabilistic reparametrisation for the decision variable, which allows optimisation of the acquisition function on a continuous space.
By employing Bayesian optimisation we incorporate uncertainty in the optimisation procedure itself, while allowing for an integrated framework in which the estimation of optimisation variables is data driven. We demonstrate the performance of our method in a simulated study.

Keywords

Status: accepted

Back to the list of papers