Don't Do What Doesn't Matter: Intrinsic Motivation with Action Usefulness

Seurin, Mathieu; Strub, Florian; Preux, Philippe; Pietquin, Olivier

Type de document :

Communication dans un congrès avec actes

Titre :

Don't Do What Doesn't Matter: Intrinsic Motivation with Action Usefulness

Auteur(s) :

Seurin, Mathieu [Auteur]
Scool [Scool]
Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Université de Lille
Centrale Lille
Strub, Florian [Auteur]
DeepMind [London]
Preux, Philippe [Auteur]
Scool [Scool]
Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Université de Lille
Centrale Lille
Pietquin, Olivier [Auteur]
Google Research [Paris]

Titre de la manifestation scientifique :

Internationnal Joint Conference on Artificial Intelligence (IJCAI)

Ville :

Montreal

Pays :

Canada

Date de début de la manifestation scientifique :

2021-08-21

Titre de l’ouvrage :

Proc. Internationnal Joint Conference on Artificial Intelligence (IJCAI)

Discipline(s) HAL :

Informatique [cs]/Apprentissage [cs.LG]

Résumé en anglais : [en]

Sparse rewards are double-edged training signals in reinforcement learning: easy to design but hard to optimize. Intrinsic motivation guidances have thus been developed toward alleviating the resulting exploration problem. ...
Lire la suite >Sparse rewards are double-edged training signals in reinforcement learning: easy to design but hard to optimize. Intrinsic motivation guidances have thus been developed toward alleviating the resulting exploration problem. They usually incentivize agents to look for new states through novelty signals. Yet, such methods encourage exhaustive exploration of the state space rather than focusing on the environment's salient interaction opportunities. We propose a new exploration method, called Don't Do What Doesn't Matter (DoWhaM), shifting the emphasis from state novelty to state with relevant actions. While most actions consistently change the state when used, e.g. moving the agent, some actions are only effective in specific states, e.g., opening a door, grabbing an object. DoWhaM detects and rewards actions that seldom affect the environment. We evaluate DoWhaM on the procedurallygenerated environment MiniGrid, against state-ofthe-art methods. Experiments consistently show that DoWhaM greatly reduces sample complexity, installing the new state-of-the-art in MiniGrid.Lire moins >

Langue :

Anglais

Comité de lecture :

Oui

Audience :

Internationale

Vulgarisation :

Non

Collections :

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

Source :

Harvested from HAL

Fichiers

https://hal.archives-ouvertes.fr/hal-03259315/document
Accès libre
Accéder au document

https://hal.archives-ouvertes.fr/hal-03259315/document
Accès libre
Accéder au document

https://hal.archives-ouvertes.fr/hal-03259315/document
Accès libre
Accéder au document

document
Accès libre
Accéder au document

Rare_Actions_Matter_IJCAI.pdf
Accès libre
Accéder au document

Don't Do What Doesn't Matter: Intrinsic ... BibTeX CSV Excel RIS

Fichiers

Don't Do What Doesn't Matter: Intrinsic ...

BibTeX

CSV

Excel

RIS