A novel graph-based estimation of the distribution algorithm and its extension using reinforcement learning

Xianneng Li, Shingo Mabu, Kotaro Hirasawa

Research output: Contribution to journalArticle

21 Citations (Scopus)

Abstract

In recent years, numerous studies have drawn the success of estimation of distribution algorithms (EDAs) to avoid the frequent breakage of building blocks of the conventional stochastic genetic operators-based evolutionary algorithms (EAs). In this paper, a novel graph-based EDA called probabilistic model building genetic network programming (PMBGNP) is proposed. Using the distinguished graph (network) structure of a graph-based EA called genetic network programming (GNP), PMBGNP ensures higher expression ability than the conventional EDAs to solve some specific problems. Furthermore, an extended algorithm called reinforced PMBGNP is proposed to combine PMBGNP and reinforcement learning to enhance the performance in terms of fitness values, search speed, and reliability. The proposed algorithms are applied to solve the problems of controlling the agents' behavior. Two problems are selected to demonstrate the effectiveness of the proposed algorithms, including the benchmark one, i.e., the Tileworld system, and a real mobile robot control.

Original languageEnglish
Article number6408015
Pages (from-to)98-113
Number of pages16
JournalIEEE Transactions on Evolutionary Computation
Volume18
Issue number1
DOIs
Publication statusPublished - 2014 Feb

Fingerprint

Network Programming
Reinforcement learning
Genetic Network
Reinforcement Learning
Genetic Programming
Probabilistic Model
Graph in graph theory
Evolutionary algorithms
Evolutionary Algorithms
Breakage
Genetic Operators
Robot Control
Network Structure
Mobile Robot
Mobile robots
Building Blocks
Fitness
Mathematical operators
Benchmark
Statistical Models

Keywords

  • Agent control
  • estimation of distribution algorithm (EDA)
  • genetic network programming (GNP)
  • graph structure
  • reinforcement learning (RL)

ASJC Scopus subject areas

  • Software
  • Computational Theory and Mathematics
  • Theoretical Computer Science

Cite this

A novel graph-based estimation of the distribution algorithm and its extension using reinforcement learning. / Li, Xianneng; Mabu, Shingo; Hirasawa, Kotaro.

In: IEEE Transactions on Evolutionary Computation, Vol. 18, No. 1, 6408015, 02.2014, p. 98-113.

Research output: Contribution to journalArticle

@article{840cfa645cba451d85cfbfbc21ed73cd,
title = "A novel graph-based estimation of the distribution algorithm and its extension using reinforcement learning",
abstract = "In recent years, numerous studies have drawn the success of estimation of distribution algorithms (EDAs) to avoid the frequent breakage of building blocks of the conventional stochastic genetic operators-based evolutionary algorithms (EAs). In this paper, a novel graph-based EDA called probabilistic model building genetic network programming (PMBGNP) is proposed. Using the distinguished graph (network) structure of a graph-based EA called genetic network programming (GNP), PMBGNP ensures higher expression ability than the conventional EDAs to solve some specific problems. Furthermore, an extended algorithm called reinforced PMBGNP is proposed to combine PMBGNP and reinforcement learning to enhance the performance in terms of fitness values, search speed, and reliability. The proposed algorithms are applied to solve the problems of controlling the agents' behavior. Two problems are selected to demonstrate the effectiveness of the proposed algorithms, including the benchmark one, i.e., the Tileworld system, and a real mobile robot control.",
keywords = "Agent control, estimation of distribution algorithm (EDA), genetic network programming (GNP), graph structure, reinforcement learning (RL)",
author = "Xianneng Li and Shingo Mabu and Kotaro Hirasawa",
year = "2014",
month = "2",
doi = "10.1109/TEVC.2013.2238240",
language = "English",
volume = "18",
pages = "98--113",
journal = "IEEE Transactions on Evolutionary Computation",
issn = "1089-778X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

TY - JOUR

T1 - A novel graph-based estimation of the distribution algorithm and its extension using reinforcement learning

AU - Li, Xianneng

AU - Mabu, Shingo

AU - Hirasawa, Kotaro

PY - 2014/2

Y1 - 2014/2

N2 - In recent years, numerous studies have drawn the success of estimation of distribution algorithms (EDAs) to avoid the frequent breakage of building blocks of the conventional stochastic genetic operators-based evolutionary algorithms (EAs). In this paper, a novel graph-based EDA called probabilistic model building genetic network programming (PMBGNP) is proposed. Using the distinguished graph (network) structure of a graph-based EA called genetic network programming (GNP), PMBGNP ensures higher expression ability than the conventional EDAs to solve some specific problems. Furthermore, an extended algorithm called reinforced PMBGNP is proposed to combine PMBGNP and reinforcement learning to enhance the performance in terms of fitness values, search speed, and reliability. The proposed algorithms are applied to solve the problems of controlling the agents' behavior. Two problems are selected to demonstrate the effectiveness of the proposed algorithms, including the benchmark one, i.e., the Tileworld system, and a real mobile robot control.

AB - In recent years, numerous studies have drawn the success of estimation of distribution algorithms (EDAs) to avoid the frequent breakage of building blocks of the conventional stochastic genetic operators-based evolutionary algorithms (EAs). In this paper, a novel graph-based EDA called probabilistic model building genetic network programming (PMBGNP) is proposed. Using the distinguished graph (network) structure of a graph-based EA called genetic network programming (GNP), PMBGNP ensures higher expression ability than the conventional EDAs to solve some specific problems. Furthermore, an extended algorithm called reinforced PMBGNP is proposed to combine PMBGNP and reinforcement learning to enhance the performance in terms of fitness values, search speed, and reliability. The proposed algorithms are applied to solve the problems of controlling the agents' behavior. Two problems are selected to demonstrate the effectiveness of the proposed algorithms, including the benchmark one, i.e., the Tileworld system, and a real mobile robot control.

KW - Agent control

KW - estimation of distribution algorithm (EDA)

KW - genetic network programming (GNP)

KW - graph structure

KW - reinforcement learning (RL)

UR - http://www.scopus.com/inward/record.url?scp=84893820282&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84893820282&partnerID=8YFLogxK

U2 - 10.1109/TEVC.2013.2238240

DO - 10.1109/TEVC.2013.2238240

M3 - Article

VL - 18

SP - 98

EP - 113

JO - IEEE Transactions on Evolutionary Computation

JF - IEEE Transactions on Evolutionary Computation

SN - 1089-778X

IS - 1

M1 - 6408015

ER -