Short-Term Scientific Mission

AutoMORK

Main theme: Algorithm Selection and Configuration
Grantee: Raul Martin Santamaria, Universidad Rey Juan Carlos, Móstoles, Spain
Host: Manuel López-Ibáñez, University of Manchester, Manchester, United Kingdom

Start date: 2024-08-22
End date: 2024-09-05
Awarded: 2024-07-31
Report approved: 2024-10-28

Description

The goals of the STSM are to enhance the automatic configuration approach used in the MORK framework by integrating support for Irace up to v3.9. Moreover, we plan to incorporate parameter ablation advice in the framework and develop a report generation tool inspired by the Iraceplot project. The plan is to reduce user intervention by automating several manual processes and identify state-of-the-art performance and evaluation techniques for arbitrary algorithm configuration. Additional activities include starting work on a black-box algorithm complexity calculator and developing a unified Facility Layout Problem solver with a common benchmark instance set.

Achievements

Significant improvements were made to the automatic configuration approach, including support for Irace up to v3.9 and the addition of parameter ablation advice in MORK. A report generation tool was developed, reducing user intervention and improving efficiency. Partial support for multi-objective optimization was added, and a black-box algorithm complexity calculator was partially implemented. Work began on a unified Facility Layout Problem solver with a common benchmark instance set. These advancements have enhanced the framework’s versatility and applicability, laying the foundation for future developments and collaborations.