Our project was very well presented at the last CP2024 (International Conference on Principles and Practice of Constraint Programming) conference held in Girona, Catalonia.

Hélène Verhaeghe (KULeuven) presented her research about the automatic learning of precedences using Graph Neural Networks in the context of RCPSP (scheduling) problems at the CP2024 conference. The presented work consists of the design of a GNN pipeline, training protocol, and a study of the usefulness of the prediction when used as additional information to the initial problem. The research shows that learned precedences mainly help to find better solutions earlier.

You can find the full paper and the slides online here and here.

Guillaume Povéda (Airbus) presented some ongoing research done in Tuples about trustworthy workforce allocation in the “The Seventh Workshop on Progress Towards the Holy Grail” workshop held during the CP2024 conference. The presented work is aiming at increasing adoption of optimisation techniques via better interactivity and explainable decision making tool.

You can find the details on the paper in the workshop webpage.