Beyond Performance: Explaining Non-Intuitive Deep Reinforcement Learning Actions in Elastic Optical Networks
Paper i proceeding, 2025
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.
Optical networks
Shapley Value for Explaining Reinforcement Learning (SVERL)
Explainable AI
Författare
Louma Mehyeddine
Universite Libanaise
Carlos Natalino Da Silva
Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk
Alaa Amro
Universite Libanaise
Jean Pierre Asdikian
Politecnico di Milano
Ihab Sbeity
Universite Libanaise
Guido Maier
Politecnico di Milano
Paolo Monti
Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk
Sebastian Troia
Politecnico di Milano
Omran Ayoub
Scuola Universitaria Professionale della Svizzera Italiana (SUPSI)
Proceedings of the 51st European Conference on Optical Communication, ECOC 2025
51st European Conference on Optical Communication
Copenhagen, Denmark,
Copenhagen, Denmark,
Efficient Confluent Edge Networks (ECO-eNET)
Europeiska kommissionen (EU) (EC/HE/101139133), 2024-01-01 -- 2028-12-31.
Styrkeområden
Informations- och kommunikationsteknik
Ämneskategorier (SSIF 2025)
Kommunikationssystem
Datavetenskap (datalogi)