Demonstration of a compositional learning framework for open and disaggregated optical network control
Résumé
We introduce an automated Compositional Learning Framework, which can dynamically combine ML models to create a composite ML service. It leverages the MLOps principle to streamline drift-aware ML workflows. We showcase its applicability in the dynamic Routing Modulation and Spectrum Allocation scenario with an open disaggregated control platform.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|