Reinforcement Learning of Intelligent Characters in Fighting Action Games
In this paper, we investigate reinforcement learning (RL) of intelligent characters, based on neural network technology, for fighting action games. RL can be either on-policy or off-policy. We apply both schemes to tabula rasa learning and adaptation. The experimental results show that (1) in tabula rasa leaning, off-policy RL outperforms on-policy RL, but (2) in adaptation, on-policy RL outperforms off-policy RL.
KeywordsReinforcement Learn Output Action Input State Action Game Static Algorithm
Unable to display preview. Download preview PDF.
- 2.Dixon, K.R., Malak, R.J., Khosla, P.K.: Incorporating prior knowledge and previously learned information into reinforcement learning agents. Tech. Rep., Institute for Complex Engineered Systems, Carnegie Mellon University (2000)Google Scholar
- 3.Cho, B.H., Jung, S.H., Seong, Y.R., Oh, H.R.: Exploiting intelligence in fighting action games using neural networks. IEICE Trans. on Information and Systems (to appear)Google Scholar
- 4.Cho, B.H., Jung, S.H., Shim, K.-H., Seong, Y.R., Oh, H.R.: Adaptation of intelligent characters to changes of game environments. In: Hao, Y., Liu, J., Wang, Y.-P., Cheung, Y.-m., Yin, H., Jiao, L., Ma, J., Jiao, Y.-C. (eds.) CIS 2005. LNCS (LNAI), vol. 3801, pp. 1064–1073. Springer, Heidelberg (2005)CrossRefGoogle Scholar