Policy gradient method using fuzzy controller in policies and its application

N. H. Noor Imanina, Harukazu Igarashi

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

Abstract

One of the reinforcement learning algorithms proposed by Igarashi and Ishihara is a combining method of policy gradient method and fuzzy control. In 2012, M. Sugimoto implemented the algorithm to the RoboCup Small Size League action decision system. The system received 30 scenes, taken from RoboCup Japan Open 2012 Competition to be learned. The purpose of this paper is to present the detailed analysis on the fuzzy rules in the policies taken from the system in order to find out the cause of the failure in the learning of 5 of the scenes received. A method was proposed to determine the rules that caused error in the learning of 5 scenes by evaluating the degree of contribution and divergence of each rule.

Original languageEnglish
Title of host publicationInternational Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT
PublisherSociety of Digital Information and Wireless Communications (SDIWC)
Pages167-174
Number of pages8
Publication statusPublished - 2014
EventInternational Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014 - Kuala Lumpur, Malaysia
Duration: 2014 Nov 172014 Nov 19

Other

OtherInternational Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014
CountryMalaysia
CityKuala Lumpur
Period14/11/1714/11/19

Fingerprint

Gradient methods
Reinforcement learning
Fuzzy rules
Fuzzy control
Learning algorithms
Controllers

Keywords

  • Action decision system
  • Fuzzy control
  • Fuzzy rules
  • Policy gradient method
  • Reinforcement learning
  • RoboCup Small Size League

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Vision and Pattern Recognition

Cite this

Noor Imanina, N. H., & Igarashi, H. (2014). Policy gradient method using fuzzy controller in policies and its application. In International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT (pp. 167-174). Society of Digital Information and Wireless Communications (SDIWC).

Policy gradient method using fuzzy controller in policies and its application. / Noor Imanina, N. H.; Igarashi, Harukazu.

International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT. Society of Digital Information and Wireless Communications (SDIWC), 2014. p. 167-174.

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

Noor Imanina, NH & Igarashi, H 2014, Policy gradient method using fuzzy controller in policies and its application. in International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT. Society of Digital Information and Wireless Communications (SDIWC), pp. 167-174, International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Kuala Lumpur, Malaysia, 14/11/17.
Noor Imanina NH, Igarashi H. Policy gradient method using fuzzy controller in policies and its application. In International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT. Society of Digital Information and Wireless Communications (SDIWC). 2014. p. 167-174
Noor Imanina, N. H. ; Igarashi, Harukazu. / Policy gradient method using fuzzy controller in policies and its application. International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT. Society of Digital Information and Wireless Communications (SDIWC), 2014. pp. 167-174
@inproceedings{1d5d71baf61c48c58190f91a8d8963fd,
title = "Policy gradient method using fuzzy controller in policies and its application",
abstract = "One of the reinforcement learning algorithms proposed by Igarashi and Ishihara is a combining method of policy gradient method and fuzzy control. In 2012, M. Sugimoto implemented the algorithm to the RoboCup Small Size League action decision system. The system received 30 scenes, taken from RoboCup Japan Open 2012 Competition to be learned. The purpose of this paper is to present the detailed analysis on the fuzzy rules in the policies taken from the system in order to find out the cause of the failure in the learning of 5 of the scenes received. A method was proposed to determine the rules that caused error in the learning of 5 scenes by evaluating the degree of contribution and divergence of each rule.",
keywords = "Action decision system, Fuzzy control, Fuzzy rules, Policy gradient method, Reinforcement learning, RoboCup Small Size League",
author = "{Noor Imanina}, {N. H.} and Harukazu Igarashi",
year = "2014",
language = "English",
pages = "167--174",
booktitle = "International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT",
publisher = "Society of Digital Information and Wireless Communications (SDIWC)",

}

TY - GEN

T1 - Policy gradient method using fuzzy controller in policies and its application

AU - Noor Imanina, N. H.

AU - Igarashi, Harukazu

PY - 2014

Y1 - 2014

N2 - One of the reinforcement learning algorithms proposed by Igarashi and Ishihara is a combining method of policy gradient method and fuzzy control. In 2012, M. Sugimoto implemented the algorithm to the RoboCup Small Size League action decision system. The system received 30 scenes, taken from RoboCup Japan Open 2012 Competition to be learned. The purpose of this paper is to present the detailed analysis on the fuzzy rules in the policies taken from the system in order to find out the cause of the failure in the learning of 5 of the scenes received. A method was proposed to determine the rules that caused error in the learning of 5 scenes by evaluating the degree of contribution and divergence of each rule.

AB - One of the reinforcement learning algorithms proposed by Igarashi and Ishihara is a combining method of policy gradient method and fuzzy control. In 2012, M. Sugimoto implemented the algorithm to the RoboCup Small Size League action decision system. The system received 30 scenes, taken from RoboCup Japan Open 2012 Competition to be learned. The purpose of this paper is to present the detailed analysis on the fuzzy rules in the policies taken from the system in order to find out the cause of the failure in the learning of 5 of the scenes received. A method was proposed to determine the rules that caused error in the learning of 5 scenes by evaluating the degree of contribution and divergence of each rule.

KW - Action decision system

KW - Fuzzy control

KW - Fuzzy rules

KW - Policy gradient method

KW - Reinforcement learning

KW - RoboCup Small Size League

UR - http://www.scopus.com/inward/record.url?scp=85019448509&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85019448509&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85019448509

SP - 167

EP - 174

BT - International Conference on Artificial Intelligence and Pattern Recognition, AIPR 2014, Held at the 3rd World Congress on Computing and Information Technology, WCIT

PB - Society of Digital Information and Wireless Communications (SDIWC)

ER -