Short-Term Scientific Mission

Automatic Configuration of Algorithms for Optimization in Logistics and Transport Problems

Main theme: Algorithm Selection and Configuration
Grantee: Imène Ait Abderrahim, Khemis Miliana University, Ain Defla, Algeria
Host: Carola Doerr, Sorbonne Université, Paris, France

Start date: 2024-09-01
End date: 2024-09-30
Awarded: 2024-07-31
Report approved: 2024-11-06

Description

Metaheuristic algorithms are traditionally developed manually, which is a labor-intensive and error-prone process that limits design exploration. In contrast, automatic algorithm configuration efficiently optimizes parameters, enabling self-adaptive algorithms to dynamically address specific problem challenges. The STSM aims to extend the automatic configuration of algorithms and black-box optimization techniques to optimize resources in logistics and transportation using automatic configuration tools like PyDGGA and irace.

Achievements

The STSM was carried out in collaboration with Dr. Carola Doerr and Dr. Adeline Fermanian, research head at the Califrais company. It focused on exploring the application of automatic algorithm configuration for solving the Delivery vehicle routing problem with time window (DVRPTW) that was identified by the Califrais company.

During the STSM, the DVRPTW from Califrais was explored and potential improvements using the automatic algorithm configuration were discussed. An initial list of parameters to be tuned in the company’s algorithm were selected and possible adaptations to the code for the automatic design algorithm to be developed were proposed. Finally, the STSM fostered future collaboration by strengthening partnerships, facilitating data exchange, and outlining plans for upcoming publications to tackle these challenges.