Playing Tetris Game with Reinforcement Learning
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This paper is implemented the architecture for solving the Tetris game which are defined as a complex problem in general through reinforcement learning. Tetris games require the actor's quick judgment ability and speed of response because the blocks must be stacked in an optimal location quickly, taking into account the shape and rotation of randomly appearing blocks. Also, since the number of cases is very large due to the various block types and order, if the subject of performance is a person, there is a limit to performance by simply relying on memory and memorization. therefore, the reinforcement learning architecture implemented in this study is applied not only to the implementation of the learning model but also We apply the Heuristic to increase the decision accuracy as the weighting method of reward. As a result, we were able to obtain high scores. Although it is not yet possible to say that he has completely conquered the tetris game, In several experiments, reinforcement learning was able to play better than some people. However, we also identified the disadvantage that heuristics are more influential on performance than learning models. In this paper, the structure of these architectures and the techniques and algorithms used are described in detail and the direction of approach is given.
Abstract
This paper is implemented the architecture for solving the Tetris game which are defined as a complex problem in general through reinforcement learning. Tetris games require the actor's quick judgment ability and speed of response because the blocks must be stacked in an optimal location quickly, taking into account the shape and rotation of randomly appearing blocks. Also, since the number of cases is very large due to the various block types and order, if the subject of performance is a person, there is a limit to performance by simply relying on memory and memorization. therefore, the reinforcement learning architecture implemented in this study is applied not only to the implementation of the learning model but also We apply the Heuristic to increase the decision accuracy as the weighting method of reward. As a result, we were able to obtain high scores. Although it is not yet possible to say that he has completely conquered the tetris game, In several experiments, reinforcement learning was able to play better than some people. However, we also identified the disadvantage that heuristics are more influential on performance than learning models. In this paper, the structure of these architectures and the techniques and algorithms used are described in detail and the direction of approach is given.
참고문헌 (19)
- Kaelbling, L. P., "Reinforcement learning : A survey", Journal of artificial intelligence research, 4, pp.237~285 (1996)
- Boyan, J. A., "Generalization in reinforcement learning : Safely approximating the value function", Advances in neural information processing systems, pp.369~376 (1995)
- Hasselt, H. V., "Double Q-learning", Advances in Neural Information Processing Systems, pp.2613~2621 (2010)
- Kulkarni, T. D., "Hierarchical deep reinforcement learning: Integrating temporal abstraction and intrinsic motivation", Advances in neural information processing systems, pp.3675~3683 (2016)
- Playing Tetris with Deep Reinforcement Learning, "Matt Stevens", Stanford University Convolutional Neural Networks, ~, pp.CS23 (2016)
- Demaine, E. D., "Tetris is hard, even to approximate", International Computing and Combinatorics Conference, pp.351~363 (2003), Springer
- Mnih. V., "Playing atari with deep reinforcement learning", arXiv preprint arXiv:1312.5602, ~ (2015)
- Wang. Z., "Dueling network architectures for deep reinforcement learning", arXiv preprint arXiv:1511.06581, ~ (2015)
- Oh. J., "Action-conditional video prediction using deep networks in atari games", Advances in neural information processing systems, pp.2863~2871 (2015)
- Mnih. V., "Asynchronous methods for deep reinforcement learning", International conference on machine learning, pp.1928~1937 (2016)
- Tijsma. D., "Comparing exploration strategies for Q-learning in random stochastic mazes", 2016 IEEE Symposium Series on Computational Intelligence, pp.1~8 (2016)
- Osband. I., "Deep exploration via bootstrapped DQN", Advances in neural information processing systems, pp.4026~4034 (2016)
- Hasselt. H. V., "Double Q-learning", Advances Neural Information Processing Systems, pp.2613~2621 (2010)
- Wiering. M. A., "Ensemble algorithms in reinforcement learning", IEEE Transactions on Systems, Man, and Cybernetics Part B, pp.930~936 (2008)
- Dayan, P., "Feudal reinforcement learning Advances", neural information processing systems, pp.271~278 (1993)
- Tesauro, G., "Temporal difference learning and TD-Gammon", Communications of the ACM, 38(3), pp.58~68 (1995)
- Aurelien Geron, "Hands-On Machine Learning with Scikit-Learn & Tensorflow" (2017), O'Reilly Media, Inc.
- Sugiyama Masashi, "Statistical Reinforcement Learning" (2015), CRC Press, Inc
- Maxim Lapan, "Deep Reinforcement Learning Hands-On" (2018), Packt Publishing, Inc