Approximate dynamic programming using fluid and diffusion approximations with applications to power management

Wei Chen, Dayu Huang, Ankur A. Kulkarni, Jayakrishnan Unnikrishnan, Quanyan Zhu, Prashant Mehta, Sean Meyn, Adam Wierman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

TD learning and its refinements are powerful tools for approximating the solution to dynamic programming problems. However, the techniques provide the approximate solution only within a prescribed finite-dimensional function class. Thus, the question that always arises is how should the function class be chosen? The goal of this paper is to propose an approach for TD learning based on choosing the function class using the solutions to associated fluid and diffusion approximations. In order to illustrate this new approach, the paper focuses on an application to dynamic speed scaling for power management.

Original languageEnglish (US)
Title of host publicationProceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009
Pages3575-3580
Number of pages6
DOIs
StatePublished - 2009
Event48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009 - Shanghai, China
Duration: Dec 15 2009Dec 18 2009

Other

Other48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009
CountryChina
CityShanghai
Period12/15/0912/18/09

Fingerprint

Approximate Dynamic Programming
Diffusion Approximation
Power Management
Dynamic programming
Fluid
Fluids
Dynamic Programming
Approximate Solution
Refinement
Scaling
Class
Power management
Learning

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Modeling and Simulation
  • Control and Optimization

Cite this

Chen, W., Huang, D., Kulkarni, A. A., Unnikrishnan, J., Zhu, Q., Mehta, P., ... Wierman, A. (2009). Approximate dynamic programming using fluid and diffusion approximations with applications to power management. In Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009 (pp. 3575-3580). [5399685] https://doi.org/10.1109/CDC.2009.5399685

Approximate dynamic programming using fluid and diffusion approximations with applications to power management. / Chen, Wei; Huang, Dayu; Kulkarni, Ankur A.; Unnikrishnan, Jayakrishnan; Zhu, Quanyan; Mehta, Prashant; Meyn, Sean; Wierman, Adam.

Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009. 2009. p. 3575-3580 5399685.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Chen, W, Huang, D, Kulkarni, AA, Unnikrishnan, J, Zhu, Q, Mehta, P, Meyn, S & Wierman, A 2009, Approximate dynamic programming using fluid and diffusion approximations with applications to power management. in Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009., 5399685, pp. 3575-3580, 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009, Shanghai, China, 12/15/09. https://doi.org/10.1109/CDC.2009.5399685
Chen W, Huang D, Kulkarni AA, Unnikrishnan J, Zhu Q, Mehta P et al. Approximate dynamic programming using fluid and diffusion approximations with applications to power management. In Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009. 2009. p. 3575-3580. 5399685 https://doi.org/10.1109/CDC.2009.5399685
Chen, Wei ; Huang, Dayu ; Kulkarni, Ankur A. ; Unnikrishnan, Jayakrishnan ; Zhu, Quanyan ; Mehta, Prashant ; Meyn, Sean ; Wierman, Adam. / Approximate dynamic programming using fluid and diffusion approximations with applications to power management. Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009. 2009. pp. 3575-3580
@inproceedings{4c4d0d749a8140ceb2eacb8dd11b5324,
title = "Approximate dynamic programming using fluid and diffusion approximations with applications to power management",
abstract = "TD learning and its refinements are powerful tools for approximating the solution to dynamic programming problems. However, the techniques provide the approximate solution only within a prescribed finite-dimensional function class. Thus, the question that always arises is how should the function class be chosen? The goal of this paper is to propose an approach for TD learning based on choosing the function class using the solutions to associated fluid and diffusion approximations. In order to illustrate this new approach, the paper focuses on an application to dynamic speed scaling for power management.",
author = "Wei Chen and Dayu Huang and Kulkarni, {Ankur A.} and Jayakrishnan Unnikrishnan and Quanyan Zhu and Prashant Mehta and Sean Meyn and Adam Wierman",
year = "2009",
doi = "10.1109/CDC.2009.5399685",
language = "English (US)",
isbn = "9781424438716",
pages = "3575--3580",
booktitle = "Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009",

}

TY - GEN

T1 - Approximate dynamic programming using fluid and diffusion approximations with applications to power management

AU - Chen, Wei

AU - Huang, Dayu

AU - Kulkarni, Ankur A.

AU - Unnikrishnan, Jayakrishnan

AU - Zhu, Quanyan

AU - Mehta, Prashant

AU - Meyn, Sean

AU - Wierman, Adam

PY - 2009

Y1 - 2009

N2 - TD learning and its refinements are powerful tools for approximating the solution to dynamic programming problems. However, the techniques provide the approximate solution only within a prescribed finite-dimensional function class. Thus, the question that always arises is how should the function class be chosen? The goal of this paper is to propose an approach for TD learning based on choosing the function class using the solutions to associated fluid and diffusion approximations. In order to illustrate this new approach, the paper focuses on an application to dynamic speed scaling for power management.

AB - TD learning and its refinements are powerful tools for approximating the solution to dynamic programming problems. However, the techniques provide the approximate solution only within a prescribed finite-dimensional function class. Thus, the question that always arises is how should the function class be chosen? The goal of this paper is to propose an approach for TD learning based on choosing the function class using the solutions to associated fluid and diffusion approximations. In order to illustrate this new approach, the paper focuses on an application to dynamic speed scaling for power management.

UR - http://www.scopus.com/inward/record.url?scp=77950814003&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77950814003&partnerID=8YFLogxK

U2 - 10.1109/CDC.2009.5399685

DO - 10.1109/CDC.2009.5399685

M3 - Conference contribution

SN - 9781424438716

SP - 3575

EP - 3580

BT - Proceedings of the 48th IEEE Conference on Decision and Control held jointly with 2009 28th Chinese Control Conference, CDC/CCC 2009

ER -