Learning to improve agent behaviours in GOAL

Dhirendra Singh, Koen V. Hindriks

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

Abstract

This paper investigates the issue of adaptability of behaviour in the context of agent-oriented programming. We focus on improving action selection in rule-based agent programming languages using a reinforcement learning mechanism under the hood. The novelty is that learning utilises the existing mental state representation of the agent, which means that (i) the programming model is unchanged and using learning within the program becomes straightforward, and (ii) adaptive behaviours can be combined with regular behaviours in a modular way. Overall, the key to effective programming in this setting is to balance between constraining behaviour using operational knowledge, and leaving flexibility to allow for ongoing adaptation. We illustrate this using different types of programs for solving the Blocks World problem.

Original languageEnglish
Title of host publicationProgramming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers
Pages158-173
Number of pages16
DOIs
Publication statusPublished - 5 Sep 2013
Externally publishedYes
Event10th International Workshop on Programming Multi-Agent Systems, ProMAS 2012 - Valencia, Spain
Duration: 5 Jun 20125 Jun 2012

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7837 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference10th International Workshop on Programming Multi-Agent Systems, ProMAS 2012
CountrySpain
CityValencia
Period5/06/125/06/12

Fingerprint

Reinforcement learning
Programming
Computer programming languages
Adaptive Behavior
Reinforcement Learning
Adaptability
Programming Model
Programming Languages
Flexibility
Learning
Context
Knowledge

Keywords

  • Agent programming
  • reinforcement learning
  • rule selection

Cite this

Singh, D., & Hindriks, K. V. (2013). Learning to improve agent behaviours in GOAL. In Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers (pp. 158-173). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7837 LNAI). https://doi.org/10.1007/978-3-642-38700-5_10
Singh, Dhirendra ; Hindriks, Koen V. / Learning to improve agent behaviours in GOAL. Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers. 2013. pp. 158-173 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{9968c7b3b0bb42dab0637dd6cadb4b57,
title = "Learning to improve agent behaviours in GOAL",
abstract = "This paper investigates the issue of adaptability of behaviour in the context of agent-oriented programming. We focus on improving action selection in rule-based agent programming languages using a reinforcement learning mechanism under the hood. The novelty is that learning utilises the existing mental state representation of the agent, which means that (i) the programming model is unchanged and using learning within the program becomes straightforward, and (ii) adaptive behaviours can be combined with regular behaviours in a modular way. Overall, the key to effective programming in this setting is to balance between constraining behaviour using operational knowledge, and leaving flexibility to allow for ongoing adaptation. We illustrate this using different types of programs for solving the Blocks World problem.",
keywords = "Agent programming, reinforcement learning, rule selection",
author = "Dhirendra Singh and Hindriks, {Koen V.}",
year = "2013",
month = "9",
day = "5",
doi = "10.1007/978-3-642-38700-5_10",
language = "English",
isbn = "9783642386992",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "158--173",
booktitle = "Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers",

}

Singh, D & Hindriks, KV 2013, Learning to improve agent behaviours in GOAL. in Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7837 LNAI, pp. 158-173, 10th International Workshop on Programming Multi-Agent Systems, ProMAS 2012, Valencia, Spain, 5/06/12. https://doi.org/10.1007/978-3-642-38700-5_10

Learning to improve agent behaviours in GOAL. / Singh, Dhirendra; Hindriks, Koen V.

Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers. 2013. p. 158-173 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7837 LNAI).

Research output: Chapter in Book / Report / Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Learning to improve agent behaviours in GOAL

AU - Singh, Dhirendra

AU - Hindriks, Koen V.

PY - 2013/9/5

Y1 - 2013/9/5

N2 - This paper investigates the issue of adaptability of behaviour in the context of agent-oriented programming. We focus on improving action selection in rule-based agent programming languages using a reinforcement learning mechanism under the hood. The novelty is that learning utilises the existing mental state representation of the agent, which means that (i) the programming model is unchanged and using learning within the program becomes straightforward, and (ii) adaptive behaviours can be combined with regular behaviours in a modular way. Overall, the key to effective programming in this setting is to balance between constraining behaviour using operational knowledge, and leaving flexibility to allow for ongoing adaptation. We illustrate this using different types of programs for solving the Blocks World problem.

AB - This paper investigates the issue of adaptability of behaviour in the context of agent-oriented programming. We focus on improving action selection in rule-based agent programming languages using a reinforcement learning mechanism under the hood. The novelty is that learning utilises the existing mental state representation of the agent, which means that (i) the programming model is unchanged and using learning within the program becomes straightforward, and (ii) adaptive behaviours can be combined with regular behaviours in a modular way. Overall, the key to effective programming in this setting is to balance between constraining behaviour using operational knowledge, and leaving flexibility to allow for ongoing adaptation. We illustrate this using different types of programs for solving the Blocks World problem.

KW - Agent programming

KW - reinforcement learning

KW - rule selection

UR - http://www.scopus.com/inward/record.url?scp=84883265063&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84883265063&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-38700-5_10

DO - 10.1007/978-3-642-38700-5_10

M3 - Conference contribution

SN - 9783642386992

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 158

EP - 173

BT - Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers

ER -

Singh D, Hindriks KV. Learning to improve agent behaviours in GOAL. In Programming Multi-Agent Systems - 10th International Workshop, ProMAS 2012, Revised Selected Papers. 2013. p. 158-173. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-38700-5_10