Asymmetric Action Abstractions for Real-Time Strategy Games
Autor(a) principal: | |
---|---|
Data de Publicação: | 2019 |
Tipo de documento: | Dissertação |
Idioma: | eng |
Título da fonte: | LOCUS Repositório Institucional da UFV |
Texto Completo: | https://locus.ufv.br//handle/123456789/31019 |
Resumo: | Action abstractions restrict the number of legal actions available for real-time plan- ning in zero-sum extensive-form games, thus allowing algorithms to focus their search on a set of promising actions. Optimal strategies derived from un-abstracted game trees are guaranteed to be no worse than optimal strategies derived from action-abstracted game trees. In practice, however, due to real-time constraints and the game tree size, one is only able to derive good strategies in un-abstracted trees in small-scale games. In this paper, we introduce an action abstraction scheme we call asymmetric abstraction. Asymmetric abstractions can retain the un-abstracted trees’ theoretical advantage over regularly abstracted trees while still allowing search algorithms to derive effective strategies, even in large-scale games. Further, asym- metric abstractions allow search algorithms to “pay more attention” in some aspects of the game by unevenly dividing the algorithm’s search effort amongst different as- pects of the game. In this paper we also introduce four algorithms that search in asymmetrically-abstracted game trees to evaluate the effectiveness of our abstrac- tion schemes. Two of those are greedy-search approaches and they are called Greedy Alpha-Beta Search and Stratified Alpha-Beta Search. The other two are versions of an algorithm we call Asymmetrically Action-Abstracted NaïveMCTS (A2N and A3N) and combine Monte Carlo Tree Search (MCTS) algorithms, naïve sampling strategy for select actions’ samples, and asymmetric abstraction scheme. In addi- tion to the search algorithms, we also introduce several strategies for generating asymmetric abstractions. An extensive set of experiments in a real-time strategy game developed for research purposes shows that search algorithms using asymmet- ric abstractions are able to outperform all other search algorithms tested. |
id |
UFV_3321fb0694f9c1220029aa74b709b613 |
---|---|
oai_identifier_str |
oai:locus.ufv.br:123456789/31019 |
network_acronym_str |
UFV |
network_name_str |
LOCUS Repositório Institucional da UFV |
repository_id_str |
2145 |
spelling |
Filho, Rubens de Oliveira Moraeshttp://lattes.cnpq.br/1137970319452572Lelis, Levi Henrique Santana de2023-06-05T19:01:16Z2023-06-05T19:01:16Z2019-02-08FILHO, Rubens de Oliveira. Asymmetric Action Abstractions for Real-Time Strategy Games. 2019. 77 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Viçosa, Viçosa. 2019.https://locus.ufv.br//handle/123456789/31019Action abstractions restrict the number of legal actions available for real-time plan- ning in zero-sum extensive-form games, thus allowing algorithms to focus their search on a set of promising actions. Optimal strategies derived from un-abstracted game trees are guaranteed to be no worse than optimal strategies derived from action-abstracted game trees. In practice, however, due to real-time constraints and the game tree size, one is only able to derive good strategies in un-abstracted trees in small-scale games. In this paper, we introduce an action abstraction scheme we call asymmetric abstraction. Asymmetric abstractions can retain the un-abstracted trees’ theoretical advantage over regularly abstracted trees while still allowing search algorithms to derive effective strategies, even in large-scale games. Further, asym- metric abstractions allow search algorithms to “pay more attention” in some aspects of the game by unevenly dividing the algorithm’s search effort amongst different as- pects of the game. In this paper we also introduce four algorithms that search in asymmetrically-abstracted game trees to evaluate the effectiveness of our abstrac- tion schemes. Two of those are greedy-search approaches and they are called Greedy Alpha-Beta Search and Stratified Alpha-Beta Search. The other two are versions of an algorithm we call Asymmetrically Action-Abstracted NaïveMCTS (A2N and A3N) and combine Monte Carlo Tree Search (MCTS) algorithms, naïve sampling strategy for select actions’ samples, and asymmetric abstraction scheme. In addi- tion to the search algorithms, we also introduce several strategies for generating asymmetric abstractions. An extensive set of experiments in a real-time strategy game developed for research purposes shows that search algorithms using asymmet- ric abstractions are able to outperform all other search algorithms tested.Coordenação de Aperfeiçoamento de Pessoal de Nível SuperiorengUniversidade Federal de ViçosaCiência da ComputaçãoProgramação (Computadores)Jogos - ProgramaçãoArquitetura de softwareAbstraçãoAlgoritmosCiência da ComputaçãoAsymmetric Action Abstractions for Real-Time Strategy Gamesinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisUniversidade Federal de ViçosaDepartamento de InformáticaMestre em Ciência da ComputaçãoViçosa - MG2019-02-08Mestradoinfo:eu-repo/semantics/openAccessreponame:LOCUS Repositório Institucional da UFVinstname:Universidade Federal de Viçosa (UFV)instacron:UFVLICENSElicense.txtlicense.txttext/plain; charset=utf-81748https://locus.ufv.br//bitstream/123456789/31019/2/license.txt8a4605be74aa9ea9d79846c1fba20a33MD52ORIGINALtexto completo.pdftexto completo.pdftexto completoapplication/pdf774677https://locus.ufv.br//bitstream/123456789/31019/1/texto%20completo.pdfaf1c25a4c4b13ab4d3ce241b2fba715fMD51123456789/310192023-06-05 16:01:17.61oai:locus.ufv.br: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Repositório InstitucionalPUBhttps://www.locus.ufv.br/oai/requestfabiojreis@ufv.bropendoar:21452023-06-05T19:01:17LOCUS Repositório Institucional da UFV - Universidade Federal de Viçosa (UFV)false |
dc.title.en.fl_str_mv |
Asymmetric Action Abstractions for Real-Time Strategy Games |
title |
Asymmetric Action Abstractions for Real-Time Strategy Games |
spellingShingle |
Asymmetric Action Abstractions for Real-Time Strategy Games Filho, Rubens de Oliveira Moraes Programação (Computadores) Jogos - Programação Arquitetura de software Abstração Algoritmos Ciência da Computação |
title_short |
Asymmetric Action Abstractions for Real-Time Strategy Games |
title_full |
Asymmetric Action Abstractions for Real-Time Strategy Games |
title_fullStr |
Asymmetric Action Abstractions for Real-Time Strategy Games |
title_full_unstemmed |
Asymmetric Action Abstractions for Real-Time Strategy Games |
title_sort |
Asymmetric Action Abstractions for Real-Time Strategy Games |
author |
Filho, Rubens de Oliveira Moraes |
author_facet |
Filho, Rubens de Oliveira Moraes |
author_role |
author |
dc.contributor.authorLattes.pt-BR.fl_str_mv |
http://lattes.cnpq.br/1137970319452572 |
dc.contributor.author.fl_str_mv |
Filho, Rubens de Oliveira Moraes |
dc.contributor.advisor1.fl_str_mv |
Lelis, Levi Henrique Santana de |
contributor_str_mv |
Lelis, Levi Henrique Santana de |
dc.subject.pt-BR.fl_str_mv |
Programação (Computadores) Jogos - Programação Arquitetura de software Abstração Algoritmos |
topic |
Programação (Computadores) Jogos - Programação Arquitetura de software Abstração Algoritmos Ciência da Computação |
dc.subject.cnpq.fl_str_mv |
Ciência da Computação |
description |
Action abstractions restrict the number of legal actions available for real-time plan- ning in zero-sum extensive-form games, thus allowing algorithms to focus their search on a set of promising actions. Optimal strategies derived from un-abstracted game trees are guaranteed to be no worse than optimal strategies derived from action-abstracted game trees. In practice, however, due to real-time constraints and the game tree size, one is only able to derive good strategies in un-abstracted trees in small-scale games. In this paper, we introduce an action abstraction scheme we call asymmetric abstraction. Asymmetric abstractions can retain the un-abstracted trees’ theoretical advantage over regularly abstracted trees while still allowing search algorithms to derive effective strategies, even in large-scale games. Further, asym- metric abstractions allow search algorithms to “pay more attention” in some aspects of the game by unevenly dividing the algorithm’s search effort amongst different as- pects of the game. In this paper we also introduce four algorithms that search in asymmetrically-abstracted game trees to evaluate the effectiveness of our abstrac- tion schemes. Two of those are greedy-search approaches and they are called Greedy Alpha-Beta Search and Stratified Alpha-Beta Search. The other two are versions of an algorithm we call Asymmetrically Action-Abstracted NaïveMCTS (A2N and A3N) and combine Monte Carlo Tree Search (MCTS) algorithms, naïve sampling strategy for select actions’ samples, and asymmetric abstraction scheme. In addi- tion to the search algorithms, we also introduce several strategies for generating asymmetric abstractions. An extensive set of experiments in a real-time strategy game developed for research purposes shows that search algorithms using asymmet- ric abstractions are able to outperform all other search algorithms tested. |
publishDate |
2019 |
dc.date.issued.fl_str_mv |
2019-02-08 |
dc.date.accessioned.fl_str_mv |
2023-06-05T19:01:16Z |
dc.date.available.fl_str_mv |
2023-06-05T19:01:16Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/masterThesis |
format |
masterThesis |
status_str |
publishedVersion |
dc.identifier.citation.fl_str_mv |
FILHO, Rubens de Oliveira. Asymmetric Action Abstractions for Real-Time Strategy Games. 2019. 77 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Viçosa, Viçosa. 2019. |
dc.identifier.uri.fl_str_mv |
https://locus.ufv.br//handle/123456789/31019 |
identifier_str_mv |
FILHO, Rubens de Oliveira. Asymmetric Action Abstractions for Real-Time Strategy Games. 2019. 77 f. Dissertação (Mestrado em Ciência da Computação) - Universidade Federal de Viçosa, Viçosa. 2019. |
url |
https://locus.ufv.br//handle/123456789/31019 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.publisher.none.fl_str_mv |
Universidade Federal de Viçosa |
dc.publisher.program.fl_str_mv |
Ciência da Computação |
publisher.none.fl_str_mv |
Universidade Federal de Viçosa |
dc.source.none.fl_str_mv |
reponame:LOCUS Repositório Institucional da UFV instname:Universidade Federal de Viçosa (UFV) instacron:UFV |
instname_str |
Universidade Federal de Viçosa (UFV) |
instacron_str |
UFV |
institution |
UFV |
reponame_str |
LOCUS Repositório Institucional da UFV |
collection |
LOCUS Repositório Institucional da UFV |
bitstream.url.fl_str_mv |
https://locus.ufv.br//bitstream/123456789/31019/2/license.txt https://locus.ufv.br//bitstream/123456789/31019/1/texto%20completo.pdf |
bitstream.checksum.fl_str_mv |
8a4605be74aa9ea9d79846c1fba20a33 af1c25a4c4b13ab4d3ce241b2fba715f |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 |
repository.name.fl_str_mv |
LOCUS Repositório Institucional da UFV - Universidade Federal de Viçosa (UFV) |
repository.mail.fl_str_mv |
fabiojreis@ufv.br |
_version_ |
1801213093286510592 |