Cartes topographiques neuronales pour l'apprentissage par renforcement sur des problèmes de contrôle non-linéaire
http://api.archives-ouvertes.fr/search/?fq=halId_s:inria-00494164&fl=title_s,en_title_s,docid,label_s,en_label_s,docType_s,authIdHal_s,halId_s,structId_i,uri_s,keyword_s,en_keyword_s,authLastNameFirstName_s,journalTitle_s,abstract_s,en_abstract_s&sort=publicationDate_s+desc
- AI
- Neural Control
- Reinforcement Learning
- Kernel Methods
We present a neural architecture which combines a new reinforcement learning algorithm with a topographic encoding of the inputs as inspired by kernel-based methods. This architecture is able to learn to control non-linear systems defined on a continuous space. Some results on a task of reaching are also given.
Institut des Sciences du Mouvement Etienne-Jules Marey | Aix Marseille Université - Tous droits réservés - Mentions légales