Reinforcement learning for problems with symmetrical restricted states

M. A.S. Kamal, Junichi Murata

Research output: Contribution to journalArticle

7 Citations (Scopus)

Abstract

A reinforcement learning method is proposed that can utilize parts of states and their partial symmetries to solve a problem efficiently. In most cases the action selection does not need considering all the states but only needs looking at parts of states or restricted state of corresponding action. Moreover, restricted states of different actions are symmetrical, and thus the action value function based on restricted states can be shared which further reduces the reinforcement learning problem size. The method is compared, in terms of simulation results and other aspects, with other standard reinforcement learning methods.

Original languageEnglish
Pages (from-to)717-727
Number of pages11
JournalRobotics and Autonomous Systems
Volume56
Issue number9
DOIs
Publication statusPublished - Sep 30 2008

Fingerprint

Reinforcement learning
Reinforcement Learning
Value Function
Partial
Symmetry
Simulation

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Software
  • Mathematics(all)
  • Computer Science Applications

Cite this

Reinforcement learning for problems with symmetrical restricted states. / Kamal, M. A.S.; Murata, Junichi.

In: Robotics and Autonomous Systems, Vol. 56, No. 9, 30.09.2008, p. 717-727.

Research output: Contribution to journalArticle

@article{75ad7a76db8e443592bb8cc4c5063f54,
title = "Reinforcement learning for problems with symmetrical restricted states",
abstract = "A reinforcement learning method is proposed that can utilize parts of states and their partial symmetries to solve a problem efficiently. In most cases the action selection does not need considering all the states but only needs looking at parts of states or restricted state of corresponding action. Moreover, restricted states of different actions are symmetrical, and thus the action value function based on restricted states can be shared which further reduces the reinforcement learning problem size. The method is compared, in terms of simulation results and other aspects, with other standard reinforcement learning methods.",
author = "Kamal, {M. A.S.} and Junichi Murata",
year = "2008",
month = "9",
day = "30",
doi = "10.1016/j.robot.2008.01.004",
language = "English",
volume = "56",
pages = "717--727",
journal = "Robotics and Autonomous Systems",
issn = "0921-8890",
publisher = "Elsevier",
number = "9",

}

TY - JOUR

T1 - Reinforcement learning for problems with symmetrical restricted states

AU - Kamal, M. A.S.

AU - Murata, Junichi

PY - 2008/9/30

Y1 - 2008/9/30

N2 - A reinforcement learning method is proposed that can utilize parts of states and their partial symmetries to solve a problem efficiently. In most cases the action selection does not need considering all the states but only needs looking at parts of states or restricted state of corresponding action. Moreover, restricted states of different actions are symmetrical, and thus the action value function based on restricted states can be shared which further reduces the reinforcement learning problem size. The method is compared, in terms of simulation results and other aspects, with other standard reinforcement learning methods.

AB - A reinforcement learning method is proposed that can utilize parts of states and their partial symmetries to solve a problem efficiently. In most cases the action selection does not need considering all the states but only needs looking at parts of states or restricted state of corresponding action. Moreover, restricted states of different actions are symmetrical, and thus the action value function based on restricted states can be shared which further reduces the reinforcement learning problem size. The method is compared, in terms of simulation results and other aspects, with other standard reinforcement learning methods.

UR - http://www.scopus.com/inward/record.url?scp=49649084490&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=49649084490&partnerID=8YFLogxK

U2 - 10.1016/j.robot.2008.01.004

DO - 10.1016/j.robot.2008.01.004

M3 - Article

VL - 56

SP - 717

EP - 727

JO - Robotics and Autonomous Systems

JF - Robotics and Autonomous Systems

SN - 0921-8890

IS - 9

ER -