Authors: Arthur Ferraz, Quentin Cappart, Thibaut Vidal

Published on: February 08, 2024

Impact Score: 8.38

Arxiv code: Arxiv:2402.06040

Summary

• What is new: Introduction of a supervised learning and optimization methodology using a graph neural network for better delivery-cost estimation in districting-and-routing problems.
• Why this is important: The challenge in minimizing expected long-term routing costs due to time-consuming evaluations of routing costs on different scenarios.
• What the research proposes: A graph neural network trained on a subset of districts to estimate delivery costs, used within an iterated local search procedure for high-quality districting plans.
• Results: Demonstrated large economic gains (10.12% on average) in five UK metropolitan areas over baseline methods, and highlighted the importance of learnable geometrical features of districts.

Technical Details

Technological frameworks used: Graph Neural Network

Models used: Supervised learning model

Data used: Districting scenarios from five metropolitan areas in the UK

Potential Impact

Logistics companies, delivery services, urban planning corporations

Want to implement this idea in a business?

We have generated a startup concept here: OptiRouteAI.