Multi-Objective Genetic Programming for Explainable Reinforcement Learning - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Multi-Objective Genetic Programming for Explainable Reinforcement Learning

Alessandro Ferreira Leite
Marc Schoenauer

Résumé

Deep reinforcement learning has met noticeable successes recently for a wide range of control problems. However, this is typically based on thousands of weights and non-linearities, making solutions complex, not easily reproducible, uninterpretable and heavy. The present paper presents genetic programming approaches for building symbolic controllers. Results are competitive, in particular in the case of delayed rewards, and the solutions are lighter by orders of magnitude and much more understandable.
Fichier principal
Vignette du fichier
_EuroGP__Multi_Objective_Genetic_Programming_for_Explainable_Reinforcement_Learning.pdf (737.31 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03886307 , version 1 (06-12-2022)

Licence

Paternité

Identifiants

Citer

Mathurin Videau, Alessandro Ferreira Leite, Olivier Teytaud, Marc Schoenauer. Multi-Objective Genetic Programming for Explainable Reinforcement Learning. EUROGP 2022 - 25th European Conference on Genetic Programming, Apr 2022, Madrid, Spain. pp.278-293, ⟨10.1007/978-3-031-02056-8_18⟩. ⟨hal-03886307⟩
98 Consultations
239 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More