Adversarially Guided Actor-Critic

Flet Berliac, Yannis; Ferret, Johan; Pietquin, Olivier; Preux, Philippe; Geist, Matthieu

Document type :

Communication dans un congrès avec actes

Title :

Adversarially Guided Actor-Critic

Author(s) :

Flet Berliac, Yannis [Auteur]
Scool [Scool]
Ferret, Johan [Auteur]
Scool [Scool]
Google Brain, Paris
Pietquin, Olivier [Auteur]

Google Brain, Paris
Preux, Philippe [Auteur]
Scool [Scool]
Geist, Matthieu [Auteur]
Google Brain, Paris

Conference title :

ICLR 2021 - International Conference on Learning Representations

City :

Vienna / Virtual

Country :

Autriche

Start date of the conference :

2021-05-04

HAL domain(s) :

Informatique [cs]/Apprentissage [cs.LG]
Statistiques [stat]/Machine Learning [stat.ML]

English abstract : [en]

Despite definite success in deep reinforcement learning problems, actor-critic algorithms are still confronted with sample inefficiency in complex environments, particularly in tasks where efficient exploration is a ...
Show more >Despite definite success in deep reinforcement learning problems, actor-critic algorithms are still confronted with sample inefficiency in complex environments, particularly in tasks where efficient exploration is a bottleneck. These methods consider a policy (the actor) and a value function (the critic) whose respective losses are built using different motivations and approaches. This paper introduces a third protagonist: the adversary. While the adversary mimics the actor by minimizing the KL-divergence between their respective action distributions, the actor, in addition to learning to solve the task, tries to differentiate itself from the adversary predictions. This novel objective stimulates the actor to follow strategies that could not have been correctly predicted from previous trajectories, making its behavior innovative in tasks where the reward is extremely rare. Our experimental analysis shows that the resulting Adversarially Guided Actor-Critic (AGAC) algorithm leads to more exhaustive exploration. Notably, AGAC outperforms current state-of-the-art methods on a set of various hard-exploration and procedurally-generated tasks.Show less >

Language :

Anglais

Peer reviewed article :

Oui

Audience :

Internationale

Popular science :

Non

Collections :

Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189

Source :

Harvested from HAL

Files

https://hal.inria.fr/hal-03167169/document
Open access
Access the document

http://arxiv.org/pdf/2102.04376
Open access
Access the document

https://hal.inria.fr/hal-03167169/document
Open access
Access the document

https://hal.inria.fr/hal-03167169/document
Open access
Access the document

Adversarially Guided Actor-Critic BibTeX CSV Excel RIS

Files

Adversarially Guided Actor-Critic

BibTeX

CSV

Excel

RIS