Beyond Performance: Explaining Non-Intuitive Deep Reinforcement Learning Actions in Elastic Optical Networks
We develop a Deep Reinforcement Learning (DRL) agent for the RMSA problem and improve the Shapley Value for Explaining Reinforcement Learning (SVERL) explainability framework by integrating policy sensitivity and feature interdependence for the RMSA problem. We then explain the proactive rejection of lightpath requests. ©2025 The Author(s)
Authors
- Sebastian Troìa (ORCID: https://orcid.org/0000-0002-4712-3767)
- Carlos Natalino (ORCID: https://orcid.org/0000-0001-7501-5547)
- Ihab Sbeity (ORCID: https://orcid.org/0000-0003-1546-3879)
- Omran Ayoub (ORCID: https://orcid.org/0000-0002-3884-3594)
- Paolo Monti (ORCID: https://orcid.org/0000-0002-5636-9910)
- Jean Pierre Asdikian
- Guido Maier (ORCID: https://orcid.org/0000-0001-6103-435X)
- Alaa Amro
- Louma Mehyeddine
Institutions
- University of Applied Sciences and Arts of Southern Switzerland (CH)
- Lebanese University (LB)
- Chalmers University of Technology (SE)
- Politecnico di Milano (IT)
Publication Details
- Published
- 2025-09-28
- DOI
- https://doi.org/10.1109/ecoc66593.2025.11263061
- Primary Topic
- Explainable Artificial Intelligence (XAI)
- Type
- article
- Field-Weighted Citation Impact
- 0.00