The Robustness of the GNN approach in application of different data-generation method

How robust is the GNN approach when applied to instances generated using different data-generation methods or distributions?

Bachelor Thesis (2026)
Author(s)

J. Lemut (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Contributor(s)

Carlos March Moya – Mentor (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Neil Yorke-Smith – Mentor (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Faculty
Electrical Engineering, Mathematics and Computer Science
More Info
expand_more
Publication Year
2026
Language
English
Graduation Date
22-06-2026
Awarding Institution
Delft University of Technology
Faculty
Electrical Engineering, Mathematics and Computer Science
Downloads counter
13
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Sea cargo transportation relies on the efficiency of ports and how effectively ships are allocated to berths. This study evaluates the robustness of the Graph Neural Network proposed by Moya et. al. [7] for the Berth Allocation Problem. It uses real world scenarios and known cases to scrutinize the agent in comparison to common heuristics and a baseline optimal solution. While the agent created schedules that were in general on par with the baseline heuristics, the agent performed particularly poorly in changes to the handling times variable.

Files

License info not available