User-centered online reinforcement learning for emergent composition of ambient applications
Résumé
User Mobility Changing needs Unpredictability Components and services Composability Unpredictable availability Environment Uncertainty Instability How to provide working and relevant composite services? Distributed learning from user feedback Announcement-based interaction protocol ARSA Multi-Agent System Decentralized decision Emergence of new composite services Environment-directed automatic service composition Dynamic and continuous adaptation to the context and the user Online reinforcement distributed learning Consideration of new appearing services User ENVIRONMENT OCE State (Sensed services)
Origine | Fichiers produits par l'(les) auteur(s) |
---|