History-dependent evaluations in POMDPs
Xavier Venel
(1, 2)
,
Bruno Ziliotto
(3, 4, 5)
Xavier Venel
- Fonction : Auteur
- PersonId : 8219
- IdHAL : xavier-venel
- ORCID : 0000-0003-1150-9139
- IdRef : 163942404
Résumé
We consider POMDPs in which the weight of the stage payoff depends on the past sequence of signals and actions occurring in the infinitely repeated problem. We prove that for all epsilon>0, there exists a strategy that is epsilon-optimal for any sequence of weights satisfying a property that interprets as "the decision-maker is patient enough". This unifies and generalizes several results of the literature, and applies notably to POMDPs with limsup payoffs.
Domaines
Optimisation et contrôle [math.OC]Format du dépôt | Notice |
---|---|
Type de dépôt | Article dans une revue |
Titre |
en
History-dependent evaluations in POMDPs
|
Résumé |
en
We consider POMDPs in which the weight of the stage payoff depends on the past sequence of signals and actions occurring in the infinitely repeated problem. We prove that for all epsilon>0, there exists a strategy that is epsilon-optimal for any sequence of weights satisfying a property that interprets as "the decision-maker is patient enough". This unifies and generalizes several results of the literature, and applies notably to POMDPs with limsup payoffs.
|
Auteur(s) |
Xavier Venel
1, 2
, Bruno Ziliotto
3, 4, 5
1
CES -
Centre d'économie de la Sorbonne
( 15080 )
- Maison des Sciences Économiques - 106-112 Boulevard de l'Hôpital - 75647 Paris Cedex 13
- France
2
PSE -
Paris School of Economics
( 301309 )
- 48 boulevard Jourdan 75014 Paris
- France
3
CEREMADE -
CEntre de REcherches en MAthématiques de la DEcision
( 60 )
- Place du Maréchal de Lattre de Tassigny 75775 - Paris Cedex 16
- France
4
CNRS -
Centre National de la Recherche Scientifique
( 441569 )
- France
5
PSL -
Université Paris Sciences et Lettres
( 564132 )
- 60 rue Mazarine 75006 Paris
- France
|
Langue du document |
Anglais
|
Date de publication électronique |
2021-04-29
|
Volume |
59
|
Numéro |
2
|
Page/Identifiant |
1730–1755
|
Classification |
AMS Subject Headings : 90C39, 90C40, 37A50, 60J20
|
Vulgarisation |
Non
|
Comité de lecture |
Oui
|
Audience |
Internationale
|
Date de publication |
2021
|
Nom de la revue |
|
Domaine(s) |
|
Mots-clés |
en
Markov decision process, partial observation, long-run average payoff
|
arXiv Id | 2004.08844 |
DOI | 10.1137/20M1332876 |
Loading...