Modular Reinforcement Learning Architectures for Artificially Intelligent Agents in Complex Game Environments

Christopher J Hanna, RJ Hickey, DK Charles, Michaela Black

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

4 Citations (Scopus)

Abstract

Abstract: Recently there has been much research focus on the use of Reinforcement Learning (RL) algorithms for game agent control. However, although it has been shown that such agents are capable of learning in real time, the high dimensionality of agent sensor state spaces still prove to be a significant barrier to progress. This paper outlines an approach to dealing with this issue by using a modular RL architecture with a fine granularity of modules. The modular approach enables a reduction of the dimensionality in complex game-like environments by dividing the state space into smaller, more manageable sub tasks. While this approach is successful in reducing dimensionality, challenges with action selection, exploration and reward allocation arise. This paper discusses approaches to overcoming these issues.keywords: {Computer architecture;Energy states;Fires;Games;Learning;Space exploration;Tiles;computer games;learning (artificial intelligence);multi-agent systems;agent sensor state spaces;artificially intelligent agents;complex game environments;game agent control;modular reinforcement learning architectures;}
LanguageEnglish
Title of host publicationUnknown Host Publication
Pages380-387
Number of pages8
DOIs
Publication statusPublished - 21 Aug 2010
EventIEEE Computational Intelligence and Games Conference - Copenhagen
Duration: 21 Aug 2010 → …

Conference

ConferenceIEEE Computational Intelligence and Games Conference
Period21/08/10 → …

Fingerprint

Intelligent agents
Reinforcement learning
Computer games
Computer architecture
Sensors
Tile
Multi agent systems
Electron energy levels
Learning algorithms
Artificial intelligence
Fires

Cite this

@inproceedings{5f40732847b9436d97b4a9d7cba0b35a,
title = "Modular Reinforcement Learning Architectures for Artificially Intelligent Agents in Complex Game Environments",
abstract = "Abstract: Recently there has been much research focus on the use of Reinforcement Learning (RL) algorithms for game agent control. However, although it has been shown that such agents are capable of learning in real time, the high dimensionality of agent sensor state spaces still prove to be a significant barrier to progress. This paper outlines an approach to dealing with this issue by using a modular RL architecture with a fine granularity of modules. The modular approach enables a reduction of the dimensionality in complex game-like environments by dividing the state space into smaller, more manageable sub tasks. While this approach is successful in reducing dimensionality, challenges with action selection, exploration and reward allocation arise. This paper discusses approaches to overcoming these issues.keywords: {Computer architecture;Energy states;Fires;Games;Learning;Space exploration;Tiles;computer games;learning (artificial intelligence);multi-agent systems;agent sensor state spaces;artificially intelligent agents;complex game environments;game agent control;modular reinforcement learning architectures;}",
author = "Hanna, {Christopher J} and RJ Hickey and DK Charles and Michaela Black",
year = "2010",
month = "8",
day = "21",
doi = "10.1109/ITW.2010.5593329",
language = "English",
pages = "380--387",
booktitle = "Unknown Host Publication",

}

Hanna, CJ, Hickey, RJ, Charles, DK & Black, M 2010, Modular Reinforcement Learning Architectures for Artificially Intelligent Agents in Complex Game Environments. in Unknown Host Publication. pp. 380-387, IEEE Computational Intelligence and Games Conference, 21/08/10. https://doi.org/10.1109/ITW.2010.5593329

Modular Reinforcement Learning Architectures for Artificially Intelligent Agents in Complex Game Environments. / Hanna, Christopher J; Hickey, RJ; Charles, DK; Black, Michaela.

Unknown Host Publication. 2010. p. 380-387.

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

TY - GEN

T1 - Modular Reinforcement Learning Architectures for Artificially Intelligent Agents in Complex Game Environments

AU - Hanna, Christopher J

AU - Hickey, RJ

AU - Charles, DK

AU - Black, Michaela

PY - 2010/8/21

Y1 - 2010/8/21

N2 - Abstract: Recently there has been much research focus on the use of Reinforcement Learning (RL) algorithms for game agent control. However, although it has been shown that such agents are capable of learning in real time, the high dimensionality of agent sensor state spaces still prove to be a significant barrier to progress. This paper outlines an approach to dealing with this issue by using a modular RL architecture with a fine granularity of modules. The modular approach enables a reduction of the dimensionality in complex game-like environments by dividing the state space into smaller, more manageable sub tasks. While this approach is successful in reducing dimensionality, challenges with action selection, exploration and reward allocation arise. This paper discusses approaches to overcoming these issues.keywords: {Computer architecture;Energy states;Fires;Games;Learning;Space exploration;Tiles;computer games;learning (artificial intelligence);multi-agent systems;agent sensor state spaces;artificially intelligent agents;complex game environments;game agent control;modular reinforcement learning architectures;}

AB - Abstract: Recently there has been much research focus on the use of Reinforcement Learning (RL) algorithms for game agent control. However, although it has been shown that such agents are capable of learning in real time, the high dimensionality of agent sensor state spaces still prove to be a significant barrier to progress. This paper outlines an approach to dealing with this issue by using a modular RL architecture with a fine granularity of modules. The modular approach enables a reduction of the dimensionality in complex game-like environments by dividing the state space into smaller, more manageable sub tasks. While this approach is successful in reducing dimensionality, challenges with action selection, exploration and reward allocation arise. This paper discusses approaches to overcoming these issues.keywords: {Computer architecture;Energy states;Fires;Games;Learning;Space exploration;Tiles;computer games;learning (artificial intelligence);multi-agent systems;agent sensor state spaces;artificially intelligent agents;complex game environments;game agent control;modular reinforcement learning architectures;}

U2 - 10.1109/ITW.2010.5593329

DO - 10.1109/ITW.2010.5593329

M3 - Conference contribution

SP - 380

EP - 387

BT - Unknown Host Publication

ER -