Genetic network programming with actor-critic and its application

Shingo Mabu; Kotaro Hirasawa; Jinglu Hu

Genetic network programming with actor-critic and its application

Shingo Mabu^*, Kotaro Hirasawa, Jinglu Hu

^*Corresponding author for this work

Graduate School of Information, Production and Systems

Research output: Contribution to conference › Paper › peer-review

Abstract

A new graph-based evolutionary algorithm named "Genetic Network Programming, GNP" was proposed. GNP represents its solutions as graph structures, which can improve the expression ability and performance. And then, GNP with Reinforcement Learning (GNP with RL) has been proposed in order to search for solutions efficiently. GNP with RL can use the current information (state and reward) and change its programs during task execution, so it has an advantage over the evolution based algorithms in case much information can be obtained during task execution. In this paper, GNP with Actor-Critic (GNP-AC) which is a new type of GNP with RL is proposed. Originally, GNP deals with discrete information (ex. right, left, etc.), but GNP with AC aims to deal with continuous information (ex. the sensor value is "32"). The proposed method is applied to the controller of the Khepera simulator and its performance is evaluated.

Original language	English
Pages	3635-3640
Number of pages	6
Publication status	Published - 2005 Dec 1
Event	SICE Annual Conference 2005 - Okayama, Japan Duration: 2005 Aug 8 → 2005 Aug 10

Conference

Conference	SICE Annual Conference 2005
Country/Territory	Japan
City	Okayama
Period	05/8/8 → 05/8/10

Keywords

Actor-Critic
Evolutionary Computation
Genetic Network Programming
Khepera robot
Reinforcement Learning

ASJC Scopus subject areas

Control and Systems Engineering
Computer Science Applications
Electrical and Electronic Engineering

Cite this

@conference{e2fb21e6920b41f09c239437dac7a909,

title = "Genetic network programming with actor-critic and its application",

abstract = "A new graph-based evolutionary algorithm named {"}Genetic Network Programming, GNP{"} was proposed. GNP represents its solutions as graph structures, which can improve the expression ability and performance. And then, GNP with Reinforcement Learning (GNP with RL) has been proposed in order to search for solutions efficiently. GNP with RL can use the current information (state and reward) and change its programs during task execution, so it has an advantage over the evolution based algorithms in case much information can be obtained during task execution. In this paper, GNP with Actor-Critic (GNP-AC) which is a new type of GNP with RL is proposed. Originally, GNP deals with discrete information (ex. right, left, etc.), but GNP with AC aims to deal with continuous information (ex. the sensor value is {"}32{"}). The proposed method is applied to the controller of the Khepera simulator and its performance is evaluated.",

keywords = "Actor-Critic, Evolutionary Computation, Genetic Network Programming, Khepera robot, Reinforcement Learning",

author = "Shingo Mabu and Kotaro Hirasawa and Jinglu Hu",

year = "2005",

month = dec,

day = "1",

language = "English",

pages = "3635--3640",

note = "SICE Annual Conference 2005 ; Conference date: 08-08-2005 Through 10-08-2005",

}

TY - CONF

T1 - Genetic network programming with actor-critic and its application

AU - Mabu, Shingo

AU - Hirasawa, Kotaro

AU - Hu, Jinglu

PY - 2005/12/1

Y1 - 2005/12/1

N2 - A new graph-based evolutionary algorithm named "Genetic Network Programming, GNP" was proposed. GNP represents its solutions as graph structures, which can improve the expression ability and performance. And then, GNP with Reinforcement Learning (GNP with RL) has been proposed in order to search for solutions efficiently. GNP with RL can use the current information (state and reward) and change its programs during task execution, so it has an advantage over the evolution based algorithms in case much information can be obtained during task execution. In this paper, GNP with Actor-Critic (GNP-AC) which is a new type of GNP with RL is proposed. Originally, GNP deals with discrete information (ex. right, left, etc.), but GNP with AC aims to deal with continuous information (ex. the sensor value is "32"). The proposed method is applied to the controller of the Khepera simulator and its performance is evaluated.

AB - A new graph-based evolutionary algorithm named "Genetic Network Programming, GNP" was proposed. GNP represents its solutions as graph structures, which can improve the expression ability and performance. And then, GNP with Reinforcement Learning (GNP with RL) has been proposed in order to search for solutions efficiently. GNP with RL can use the current information (state and reward) and change its programs during task execution, so it has an advantage over the evolution based algorithms in case much information can be obtained during task execution. In this paper, GNP with Actor-Critic (GNP-AC) which is a new type of GNP with RL is proposed. Originally, GNP deals with discrete information (ex. right, left, etc.), but GNP with AC aims to deal with continuous information (ex. the sensor value is "32"). The proposed method is applied to the controller of the Khepera simulator and its performance is evaluated.

KW - Actor-Critic

KW - Evolutionary Computation

KW - Genetic Network Programming

KW - Khepera robot

KW - Reinforcement Learning

UR - http://www.scopus.com/inward/record.url?scp=33645283613&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33645283613&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:33645283613

SP - 3635

EP - 3640

T2 - SICE Annual Conference 2005

Y2 - 8 August 2005 through 10 August 2005

ER -

Genetic network programming with actor-critic and its application

Abstract

Conference

Keywords

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this