Efficient Learning and Planning Within the Dyna Framework

Jing Peng, Ronald J. Williams

Research output: Contribution to journalArticleResearchpeer-review

66 Citations (Scopus)

Abstract

Sutton’s Dyna framework provides a novel and computationally appealing way to integrate learning, planning, and reacting in autonomous agents. Examined here is a class of strategies designed to enhance the learning and planning power of Dyna systems by increasing their computational efficiency. The benefit of using these strategies is demonstrated on some simple abstract learning tasks.

Original languageEnglish
Pages (from-to)437-454
Number of pages18
JournalAdaptive Behavior
Volume1
Issue number4
DOIs
StatePublished - 1 Jan 1993

Fingerprint

Learning
Efficiency
Power (Psychology)

Keywords

  • dynamic programming
  • reinforcement learning
  • sequential decision problems

Cite this

Peng, Jing ; Williams, Ronald J. / Efficient Learning and Planning Within the Dyna Framework. In: Adaptive Behavior. 1993 ; Vol. 1, No. 4. pp. 437-454.
@article{fecc5ae2602f44d9853f14b5842cf5ce,
title = "Efficient Learning and Planning Within the Dyna Framework",
abstract = "Sutton’s Dyna framework provides a novel and computationally appealing way to integrate learning, planning, and reacting in autonomous agents. Examined here is a class of strategies designed to enhance the learning and planning power of Dyna systems by increasing their computational efficiency. The benefit of using these strategies is demonstrated on some simple abstract learning tasks.",
keywords = "dynamic programming, reinforcement learning, sequential decision problems",
author = "Jing Peng and Williams, {Ronald J.}",
year = "1993",
month = "1",
day = "1",
doi = "10.1177/105971239300100403",
language = "English",
volume = "1",
pages = "437--454",
journal = "Adaptive Behavior",
issn = "1059-7123",
publisher = "SAGE Publications Ltd",
number = "4",

}

Efficient Learning and Planning Within the Dyna Framework. / Peng, Jing; Williams, Ronald J.

In: Adaptive Behavior, Vol. 1, No. 4, 01.01.1993, p. 437-454.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Efficient Learning and Planning Within the Dyna Framework

AU - Peng, Jing

AU - Williams, Ronald J.

PY - 1993/1/1

Y1 - 1993/1/1

N2 - Sutton’s Dyna framework provides a novel and computationally appealing way to integrate learning, planning, and reacting in autonomous agents. Examined here is a class of strategies designed to enhance the learning and planning power of Dyna systems by increasing their computational efficiency. The benefit of using these strategies is demonstrated on some simple abstract learning tasks.

AB - Sutton’s Dyna framework provides a novel and computationally appealing way to integrate learning, planning, and reacting in autonomous agents. Examined here is a class of strategies designed to enhance the learning and planning power of Dyna systems by increasing their computational efficiency. The benefit of using these strategies is demonstrated on some simple abstract learning tasks.

KW - dynamic programming

KW - reinforcement learning

KW - sequential decision problems

UR - http://www.scopus.com/inward/record.url?scp=84977063352&partnerID=8YFLogxK

U2 - 10.1177/105971239300100403

DO - 10.1177/105971239300100403

M3 - Article

VL - 1

SP - 437

EP - 454

JO - Adaptive Behavior

JF - Adaptive Behavior

SN - 1059-7123

IS - 4

ER -