Budgeted Reinforcement Learning in Continuous State Space - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

Budgeted Reinforcement Learning in Continuous State Space

Résumé

A Budgeted Markov Decision Process (BMDP) is an extension of a Markov Decision Process to critical applications requiring safety constraints. It relies on a notion of risk implemented in the shape of a cost signal constrained to lie below an-adjustable-threshold. So far, BMDPs could only be solved in the case of finite state spaces with known dynamics. This work extends the state-of-the-art to continuous spaces environments and unknown dynamics. We show that the solution to a BMDP is a fixed point of a novel Budgeted Bellman Optimality operator. This observation allows us to introduce natural extensions of Deep Reinforcement Learning algorithms to address large-scale BMDPs. We validate our approach on two simulated applications: spoken dialogue and autonomous driving.
Fichier principal
Vignette du fichier
9128-budgeted-reinforcement-learning-in-continuous-state-space.pdf (524.04 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02375727 , version 1 (22-11-2019)

Identifiants

Citer

Nicolas Carrara, Edouard Leurent, Romain Laroche, Tanguy Urvoy, Odalric-Ambrym Maillard, et al.. Budgeted Reinforcement Learning in Continuous State Space. Conference on Neural Information Processing Systems, Dec 2019, Vancouver, Canada. ⟨hal-02375727⟩
109 Consultations
144 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More