Data-driven approximate dynamic programming : A linear programming approach

Sutter, Tobias; Kamoutsi, Angeliki; Esfahani, Peyman Mohajerin; Lygeros, John

Data-driven approximate dynamic programming : A linear programming approach

Dateien

Zu diesem Dokument gibt es keine Dateien.

Datum

2017

Autor:innen

Sutter, Tobias

Kamoutsi, Angeliki

Esfahani, Peyman Mohajerin

Lygeros, John

DOI (zitierfähiger Link)

https://doi.org/10.1109/CDC.2017.8264426

Sammlungen

Informatik und Informationswissenschaft

Publikationstyp

Beitrag zu einem Konferenzband

Publikationsstatus

Published

Erschienen in

2017 IEEE 56th Annual Conference on Decision and Control (CDC). Piscataway, NJ: IEEE, 2017, pp. 5174-5179. ISBN 978-1-5090-2873-3. Available under: doi: 10.1109/CDC.2017.8264426

Zusammenfassung

This article presents an approximation scheme for the infinite-dimensional linear programming formulation of discrete-time Markov control processes via a finite-dimensional convex program, when the dynamics are unknown and learned from data. We derive a probabilistic explicit error bound between the data-driven finite convex program and the original infinite linear program. We further discuss the sample complexity of the error bound which translates to the number of samples required for an a priori approximation accuracy. Our analysis sheds light on the impact of the choice of basis functions for approximating the true value function. Finally, the relevance of the method is illustrated on a truncated LQG problem.

Fachgebiet (DDC)

004 Informatik

Konferenz

IEEE 56th Annual Conference on Decision and Control (CDC), 12. Dez. 2017 - 15. Dez. 2017, Melbourne, Australia

Zitieren

ISO 690

SUTTER, Tobias, Angeliki KAMOUTSI, Peyman Mohajerin ESFAHANI, John LYGEROS, 2017. Data-driven approximate dynamic programming : A linear programming approach. IEEE 56th Annual Conference on Decision and Control (CDC). Melbourne, Australia, 12. Dez. 2017 - 15. Dez. 2017. In: 2017 IEEE 56th Annual Conference on Decision and Control (CDC). Piscataway, NJ: IEEE, 2017, pp. 5174-5179. ISBN 978-1-5090-2873-3. Available under: doi: 10.1109/CDC.2017.8264426

BibTex

@inproceedings{Sutter2017Datad-55738,
  year={2017},
  doi={10.1109/CDC.2017.8264426},
  title={Data-driven approximate dynamic programming : A linear programming approach},
  isbn={978-1-5090-2873-3},
  publisher={IEEE},
  address={Piscataway, NJ},
  booktitle={2017 IEEE 56th Annual Conference on Decision and Control (CDC)},
  pages={5174--5179},
  author={Sutter, Tobias and Kamoutsi, Angeliki and Esfahani, Peyman Mohajerin and Lygeros, John}
}

RDF

<rdf:RDF
    xmlns:dcterms="http://purl.org/dc/terms/"
    xmlns:dc="http://purl.org/dc/elements/1.1/"
    xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
    xmlns:bibo="http://purl.org/ontology/bibo/"
    xmlns:dspace="http://digital-repositories.org/ontologies/dspace/0.1.0#"
    xmlns:foaf="http://xmlns.com/foaf/0.1/"
    xmlns:void="http://rdfs.org/ns/void#"
    xmlns:xsd="http://www.w3.org/2001/XMLSchema#" > 
  <rdf:Description rdf:about="https://kops.uni-konstanz.de/server/rdf/resource/123456789/55738">
    <dc:contributor>Sutter, Tobias</dc:contributor>
    <dspace:isPartOfCollection rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <dc:contributor>Esfahani, Peyman Mohajerin</dc:contributor>
    <dcterms:available rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2021-12-02T12:46:29Z</dcterms:available>
    <dcterms:isPartOf rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <dcterms:rights rdf:resource="https://rightsstatements.org/page/InC/1.0/"/>
    <dc:creator>Lygeros, John</dc:creator>
    <bibo:uri rdf:resource="https://kops.uni-konstanz.de/handle/123456789/55738"/>
    <dc:creator>Kamoutsi, Angeliki</dc:creator>
    <dc:creator>Esfahani, Peyman Mohajerin</dc:creator>
    <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2021-12-02T12:46:29Z</dc:date>
    <dc:language>eng</dc:language>
    <dcterms:issued>2017</dcterms:issued>
    <foaf:homepage rdf:resource="http://localhost:8080/"/>
    <dc:creator>Sutter, Tobias</dc:creator>
    <dc:rights>terms-of-use</dc:rights>
    <void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/>
    <dcterms:abstract xml:lang="eng">This article presents an approximation scheme for the infinite-dimensional linear programming formulation of discrete-time Markov control processes via a finite-dimensional convex program, when the dynamics are unknown and learned from data. We derive a probabilistic explicit error bound between the data-driven finite convex program and the original infinite linear program. We further discuss the sample complexity of the error bound which translates to the number of samples required for an a priori approximation accuracy. Our analysis sheds light on the impact of the choice of basis functions for approximating the true value function. Finally, the relevance of the method is illustrated on a truncated LQG problem.</dcterms:abstract>
    <dc:contributor>Lygeros, John</dc:contributor>
    <dc:contributor>Kamoutsi, Angeliki</dc:contributor>
    <dcterms:title>Data-driven approximate dynamic programming : A linear programming approach</dcterms:title>
  </rdf:Description>
</rdf:RDF>

Universitätsbibliographie

Nein