LIEF: Learning to Influence through Evaluative Feedback - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

LIEF: Learning to Influence through Evaluative Feedback

Résumé

We present a multi-agent reinforcement learning framework where rewards are not only generated by the environment but also by other peers in it through inter-agent evaluative feedback. We show that our method allows agents to effectively learn how to use this feedback channel to influence their peers' goals and move from independent or conflicting objectives to more coinciding and concurring ones. We advance with this in the field of cooperative reinforcement learning; not by aiding agents in increasing performance in an already cooperative environment, but by giving them a tool to transform antagonistic environments into cooperative ones.
Fichier principal
Vignette du fichier
ALA2021_paper_19.pdf (1.16 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03224579 , version 1 (11-05-2021)

Identifiants

  • HAL Id : hal-03224579 , version 1

Citer

Ramona Merhej, Mohamed Chetouani. LIEF: Learning to Influence through Evaluative Feedback. Adaptive and Learning Agents Workshop (AAMAS 2021), May 2021, London (virtual), United Kingdom. ⟨hal-03224579⟩
154 Consultations
120 Téléchargements

Partager

Gmail Facebook X LinkedIn More