Reinforcement Learning para treino do Pac-Man em Speedrun / Reinforcement Learning for Pac-Man Speedrun Training
DOI:
https://doi.org/10.34117/bjdv5n11-242Keywords:
Reinforcement Learning, Games, Q-learning, Speedrun, Machine Learning, Pac-Man.Abstract
Os algoritmos de Reinforcement Learning têm o objetivo de melhorar o comportamento do agente em um ambiente específico. Esses algoritmos se mostraram excelentes na melhoria da inteligência artificial em games de diversos gêneros. Nesse trabalho é usado o algoritmo Q-Learning do Reinforcement Learning aplicado no jogo Pac-Man, um dos jogos mais tradicionais que fomentam o desenvolvimento de novos algoritmos para a área de inteligência artificial. O objetivo principal é treinar o Pac-Man para escolher os melhores caminhos e assim terminar o jogo o mais rápido possível, de acordo com o modo de jogo speedrun. Os resultados obtidos por esse trabalho mostraram que foi possível fazer o Pac-Man ter bons resultados no modo speedrun.
References
A. L. Samuel, “Some Studies in Machine Learning Using the Game of Checkers,” IBM Journal of Research and Development, vol. 3, no. 3, pp. 210–229, July 1959.
R. Haridy. The year AI beat us at all our own games. https://newatlas:com/ai-2017-beating-humans-games/52741/. Acesso em: 2018-03-29.
G. Narula. Machine Learning in Gaming – Building AIs to Conquer Virtual Worlds. https://www:techemergence:com/machine-learning-in-gaming-building-ais-to-conquer-virtual-worlds/. Acesso em: 2018-03-29.
D. B. Ortega. Machine Learning Applied to Pac-Man Final Report. https://upcommons:upc:edu/bitstream/handle/2099:1/26448/108745:pdf. Acesso em: 2018-06-30.
P. Chase. Pac-Man Leaderboard. https://www:speedrun:com/Pac-Man. Acesso em: 2018-03-29.
M. do Computador. História do Ms. Pac-Man.. http://www:museudocomputador:org:br/historia-pacman. Acesso em: 2018-07-15.
C. Birch. Understanding Pac-Man Ghost Behavior. http://gameinternals:com/post/2072558330/understanding-pac-man-ghost-behavior. Acesso em: 2018-08-20.
R. Scully-Blaker. (2014, 01) A practiced practice: Speedrunning through space with de certeau and virilio. http://gamestudies:org/1401/articles/scullyblaker. Acesso em: 2018-04-25.
M. Uyama. Speed Demos Archive. http://speeddemosarchive:com/. Acesso em: 2018-02-23.
——. Games Done Quick. https://gamesdonequick:com/. Acesso em: 2018-02-23.
B. Bertoli. How To Start Speedrunning Video Games. https://kotaku:com/how-to-start-speedrunning-video-games-1796984207. Acesso em: 2018-07-21.
T. Matiisen. Demystifying deep reinforcement learning. http://neuro:cs:ut:ee/demystifying-deep-reinforcement-learning/. Acesso em: 2018-07-17.
Adelikat. TAS(Tool Assisted Speedrun). http://tasvideos:org. Acesso em: 2018-07-21.
R. S. Sutton and A. G. Barto, Introduction to Reinforcement Learning, 1st ed. Cambridge, MA, USA: MIT Press, 1998.
R. v. LOON. Machine learning explained: Understanding supervised, unsupervised, and reinforcement learning. http://bigdata-madesimple:com/machine-learning-explained-understanding-supervised-unsupervised-and-reinforcementlearning/. Acesso em: 2018-03-12.
P. Dayan and Y. Niv, “Reinforcement learning: The Good, The Bad and The Ugly,” vol. 18, pp. 185–96, 09 2008.
M. Ashraf. Reinforcement Learning Demystified: A Gentle Introduction. https://towardsdatascience:com/reinforcement-learning-demystified-36c39c11ec14. Acesso em: 2018-07-25.
T. v. d. Ouderaa. Deep Reinforcement Learning in Pac-man. https://esc:fnwi:uva:nl/thesis/centraal/files/f323981448:pdf. Acesso em: 2018-07-18.
U. P. de Catalunya. PACMAN and reinforcement learning. http://www:lsi:upc:edu/~bejar/apren/docum/trans/pacman-eng:pdf. Acesso em: 2018-07-17.
L. Bom, R. Henken, and M. Wiering, “Reinforcement learning to train ms. pac-man using higher-order action-relative inputs,” in 2013 IEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL), April 2013, pp. 156–163.
G. Lample and D. S. Chaplot, “Playing FPS Games with Deep Reinforcement Learning,” in AAAI Conference on Artificial Intelligence, 2017.
C. J. Watkins and P. Dayan, “Q-learning,” Machine learning, vol. 8, no. 3-4, pp. 279–292, 1992.
J. McCullock. Mnemosyne studio - q-learning. http://mnemstudio:org/path-finding-q-learning:htm. Acesso em: 2018-07-25.
M. Alzantot. Deep Reinforcement Learning Demystified (Episode 0). https://medium:com/@m:alzantot/deep-reinforcement-learning-demystified-episode-02198c05a6124. Acesso em: 2018-07-25.
O. S. University. PAC MAN as Markov Decision Process. https://oregonstate:instructure:com/files/67193237/download?downloadfrd=1. Acesso em: 2018-07-25.
K. V. P. J. Svozil, Daniel. Introduction to multi-layer feed-forward neural networks. https://pdfs:semanticscholar:org/09cf/727949c915ae8fd8b0a2ead55bcafedc0a12:pdf. Acesso em: 2018-08-15.
A. Oliveira, A. Souza, W. Lacerda, and L. Resende Gonçalves, “Aplicação de redes neurais artificiais na previsão da produção de álcool,” Ciencia E Agrotecnologia - CIENC AGROTEC, vol. 34, 04 2010.
Saurabh. Backpropagation – Algorithm For Training A Neural Network. https://www:edureka:co/blog/backpropagation/. Acesso em: 2018-09-17.
Langbroek, David, Aurucci, Davide, Roca, A. Albert Segarra, and Fthi. Learning to play Pac-Man with Reinforcement Learning. https://github:com/albertsgrc/q-mspacman. Acesso em: 2018-06-15.
F. Estévez, A. Ruiz, and P. Martin. Training Pac-Man bots using Reinforcement Learning and Case-based Reasoning. http://ceur-ws:org/Vol-1957/CoSeCiVi17paper14:pdf. Acesso em: 2018-07-25.
R. Niel, J. Krebbers, M. Drugan, and M. Wiering, “Hierarchical reinforcement learning for real-time strategy games,” in International Conference on Agents and Artificial Intelligence, 01 2018.
N. Tziortziotis, K. Tziortziotis, and K. Blekas, “Play ms. pac-man using an advanced reinforcement learning agent,” in Artificial Intelligence: Methods and Applications, A. Likas, K. Blekas, and D. Kalles, Eds. Cham: Springer International Publishing, 2014, pp. 71–83.
