DISCORL: Continual reinforcement learning via policy distillation - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

DISCORL: Continual reinforcement learning via policy distillation

Résumé

In multi-task reinforcement learning there are two main challenges: at training time, the ability to learn different policies with a single model; at test time, inferring which of those policies applying without an external signal. In the case of continual reinforcement learning a third challenge arises: learning tasks sequentially without forgetting the previous ones. In this paper, we tackle these challenges by proposing DisCoRL, an approach combining state representation learning and policy distillation. We experiment on a sequence of three simulated 2D navigation tasks with a 3 wheel omni-directional robot. Moreover, we tested our approach's robustness by transferring the final policy into a real life setting. The policy can solve all tasks and automatically infer which one to run.
Fichier principal
Vignette du fichier
NeurIPS19 DRL ws- DISCORL- CONTINUAL REINFORCEMENT LEARNING VIA POLICY DISTILLATION.pdf (1.85 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02381494 , version 1 (26-11-2019)

Identifiants

  • HAL Id : hal-02381494 , version 1

Citer

René Traoré, Hugo Caselles-Dupré, Timothée Lesort, Te Sun, Guanghang Cai, et al.. DISCORL: Continual reinforcement learning via policy distillation: A preprint. NeurIPS workshop on Deep Reinforcement Learning, Dec 2019, Vancouver, Canada. ⟨hal-02381494⟩
122 Consultations
156 Téléchargements

Partager

Gmail Facebook X LinkedIn More