CoyPu: Cognitive Economy Intelligence Platform for the Resillience of Economic Ecosystems
| Leitung: | Prof. Marius Lindauer and Prof. Maria Esther-Vidal (L3S/LUH) |
| Team: | InfAI, DATEV eg., eccenca GmbH, Implisense GmbH, Deutsches Institut für Wirtschaftsforschung, Leibniz Informationszentrum Technik und Naturwissenschaften, Hamburger Informatik Technologie-Center e.V., Selbstregulierung Informationswirtschaft e.V., Infineo |
| Jahr: | 2021 |
| Förderung: | Innovationswettbewerb Künstliche Intelligenz (BMWK) |
| Laufzeit: | 2021-2024 |
Our Contribution
There are multiple contributions that we offer;
1. We investigate (efficient) hyperparameter tuning for graph-based models, levaraging a new type of fidelity; i.e. low cost approximation of the model's performance. In particular, we seek to find the subgraph the model - conditioned on its hyperaparameters - attends to. Using these important parts of the graph for premature training of another model yield training efficiency, while providing accurate performance estimates.
2. We develop a spatio-temporal graph model capable of predicting when parts of the supply chain are disrupted. The model is aimed at predicting the propagation of events through the supply-chain.
3. Offer our project partners help in automating their hyperparameter tuning
