Orthogonalized dynamic programming state space for efficient value function approximation

Bancha Ariyajunya, Victoria C P Chen, Seoung Bum Kim

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

1 Citation (Scopus)

Abstract

Dynamic programming (DP) is a mathematical programming method for optimizing a system changing over time and has been used to solve multi-stage optimization problems in manufacturing systems, environmental engineering, and many other fields. Exact solutions are only possible for small problems or under very limiting restrictions. Given recent advances in computational power, approximate DP (ADP) methods now exist; however, they are still subject to the "curse of dimensionality" and rendered computationally intractable in high-dimensions, with few exceptions. In addition, most continuous-state problems require an approximate solution through discretization of the state space. By incorporating a design and analysis of computer experiments (DACE) approach, which uses experimental design and statistical modeling to efficiently represent computer experiment output, computationallytractable ADP methods for continuous-state problems are possible. However, ideal experimental designs are orthogonal, and when the state variables are correlated, ideal experimental designs will not appropriately represent the state space. Data mining methods are employed in this study for two purposes: (1) to orthogonalize a DP state space and enable the use of ideal experimental designs, and (2) to reduce the dimensionality of a DP problem. Results are presented for an Atlanta ozone pollution problem.

Original languageEnglish
Title of host publicationIIE Annual Conference and Expo 2010 Proceedings
PublisherInstitute of Industrial Engineers
Publication statusPublished - 2010 Jan 1
EventIIE Annual Conference and Expo 2010 - Cancun, Mexico
Duration: 2010 Jun 52010 Jun 9

Other

OtherIIE Annual Conference and Expo 2010
CountryMexico
CityCancun
Period10/6/510/6/9

Fingerprint

Dynamic programming
Design of experiments
Environmental engineering
Mathematical programming
Ozone
Data mining
Pollution
Experiments

Keywords

  • Approximate dynamic programming
  • Data mining
  • Design and analysis of computer experiments
  • Ozone pollution

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering

Cite this

Ariyajunya, B., Chen, V. C. P., & Kim, S. B. (2010). Orthogonalized dynamic programming state space for efficient value function approximation. In IIE Annual Conference and Expo 2010 Proceedings Institute of Industrial Engineers.

Orthogonalized dynamic programming state space for efficient value function approximation. / Ariyajunya, Bancha; Chen, Victoria C P; Kim, Seoung Bum.

IIE Annual Conference and Expo 2010 Proceedings. Institute of Industrial Engineers, 2010.

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

Ariyajunya, B, Chen, VCP & Kim, SB 2010, Orthogonalized dynamic programming state space for efficient value function approximation. in IIE Annual Conference and Expo 2010 Proceedings. Institute of Industrial Engineers, IIE Annual Conference and Expo 2010, Cancun, Mexico, 10/6/5.
Ariyajunya B, Chen VCP, Kim SB. Orthogonalized dynamic programming state space for efficient value function approximation. In IIE Annual Conference and Expo 2010 Proceedings. Institute of Industrial Engineers. 2010
Ariyajunya, Bancha ; Chen, Victoria C P ; Kim, Seoung Bum. / Orthogonalized dynamic programming state space for efficient value function approximation. IIE Annual Conference and Expo 2010 Proceedings. Institute of Industrial Engineers, 2010.
@inproceedings{a1a39712a84a4a0d8b5656d932c61b29,
title = "Orthogonalized dynamic programming state space for efficient value function approximation",
abstract = "Dynamic programming (DP) is a mathematical programming method for optimizing a system changing over time and has been used to solve multi-stage optimization problems in manufacturing systems, environmental engineering, and many other fields. Exact solutions are only possible for small problems or under very limiting restrictions. Given recent advances in computational power, approximate DP (ADP) methods now exist; however, they are still subject to the {"}curse of dimensionality{"} and rendered computationally intractable in high-dimensions, with few exceptions. In addition, most continuous-state problems require an approximate solution through discretization of the state space. By incorporating a design and analysis of computer experiments (DACE) approach, which uses experimental design and statistical modeling to efficiently represent computer experiment output, computationallytractable ADP methods for continuous-state problems are possible. However, ideal experimental designs are orthogonal, and when the state variables are correlated, ideal experimental designs will not appropriately represent the state space. Data mining methods are employed in this study for two purposes: (1) to orthogonalize a DP state space and enable the use of ideal experimental designs, and (2) to reduce the dimensionality of a DP problem. Results are presented for an Atlanta ozone pollution problem.",
keywords = "Approximate dynamic programming, Data mining, Design and analysis of computer experiments, Ozone pollution",
author = "Bancha Ariyajunya and Chen, {Victoria C P} and Kim, {Seoung Bum}",
year = "2010",
month = "1",
day = "1",
language = "English",
booktitle = "IIE Annual Conference and Expo 2010 Proceedings",
publisher = "Institute of Industrial Engineers",

}

TY - GEN

T1 - Orthogonalized dynamic programming state space for efficient value function approximation

AU - Ariyajunya, Bancha

AU - Chen, Victoria C P

AU - Kim, Seoung Bum

PY - 2010/1/1

Y1 - 2010/1/1

N2 - Dynamic programming (DP) is a mathematical programming method for optimizing a system changing over time and has been used to solve multi-stage optimization problems in manufacturing systems, environmental engineering, and many other fields. Exact solutions are only possible for small problems or under very limiting restrictions. Given recent advances in computational power, approximate DP (ADP) methods now exist; however, they are still subject to the "curse of dimensionality" and rendered computationally intractable in high-dimensions, with few exceptions. In addition, most continuous-state problems require an approximate solution through discretization of the state space. By incorporating a design and analysis of computer experiments (DACE) approach, which uses experimental design and statistical modeling to efficiently represent computer experiment output, computationallytractable ADP methods for continuous-state problems are possible. However, ideal experimental designs are orthogonal, and when the state variables are correlated, ideal experimental designs will not appropriately represent the state space. Data mining methods are employed in this study for two purposes: (1) to orthogonalize a DP state space and enable the use of ideal experimental designs, and (2) to reduce the dimensionality of a DP problem. Results are presented for an Atlanta ozone pollution problem.

AB - Dynamic programming (DP) is a mathematical programming method for optimizing a system changing over time and has been used to solve multi-stage optimization problems in manufacturing systems, environmental engineering, and many other fields. Exact solutions are only possible for small problems or under very limiting restrictions. Given recent advances in computational power, approximate DP (ADP) methods now exist; however, they are still subject to the "curse of dimensionality" and rendered computationally intractable in high-dimensions, with few exceptions. In addition, most continuous-state problems require an approximate solution through discretization of the state space. By incorporating a design and analysis of computer experiments (DACE) approach, which uses experimental design and statistical modeling to efficiently represent computer experiment output, computationallytractable ADP methods for continuous-state problems are possible. However, ideal experimental designs are orthogonal, and when the state variables are correlated, ideal experimental designs will not appropriately represent the state space. Data mining methods are employed in this study for two purposes: (1) to orthogonalize a DP state space and enable the use of ideal experimental designs, and (2) to reduce the dimensionality of a DP problem. Results are presented for an Atlanta ozone pollution problem.

KW - Approximate dynamic programming

KW - Data mining

KW - Design and analysis of computer experiments

KW - Ozone pollution

UR - http://www.scopus.com/inward/record.url?scp=84901049854&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901049854&partnerID=8YFLogxK

M3 - Conference contribution

BT - IIE Annual Conference and Expo 2010 Proceedings

PB - Institute of Industrial Engineers

ER -