Multiagent reinforcement learning with the partly high-dimensional state space

被引：4
作者：
Department of Electrical and Computer Engineering, Nagoya Institute of Technology, Nagoya, 466-8555, Japan [1 ]
机构：
[1] Department of Electrical and Computer Engineering, Nagoya Institute of Technology, Nagoya
来源：
Syst Comput Jpn | 2006年 / 9卷 / 22-31期
关键词：
Modular Q-learning; Multiagent; Reinforcement learning; State space;
D O I：
10.1002/scj.20526
中图分类号：
学科分类号：
摘要：
One method of designing a multiagent system is called multiagent reinforcement learning. In multiagent reinforcement learning, an agent also observes the other agents as part of the environment. As a result, as the number of agents increases, the state space increases exponentially (curse of dimensionality), and the learning speed decreases dramatically. The amount of memory required for learning also becomes enormous. Modular Q-learning, which was proposed as a technique for solving this problem, has the disadvantage that the learning performance decreases due to the incompleteness of perception. In the current research, the authors propose the HMQL technique for improving the learning performance of Modular Q-learning by using a method of partially increasing the dimensionality of the state space. © 2006 Wiley Periodicals, Inc.
引用
收藏
页码：22 / 31
页数：9
相关论文
共 14 条
  • [1] Sutton R.S., Barto A.G., Reinforcement Learning: An Introduction, (1998)
  • [2] Arai S., Multiagent reinforcement learning - Uniting planning, theory, and various techniques for implementation, J Japan Soc Artif Intell, 16, pp. 476-481, (2001)
  • [3] Crites R.H., Barto A.G., Improving elevator performance using reinforcement learning, Adv Neural Inf Process Syst, 8, pp. 1017-1023, (1996)
  • [4] Wolpert D.H., Sill J., Turner K., Reinforcement learning in distributed domains: Beyond team games, Proc 17th National Conference on Artificial Intelligence, pp. 819-824, (2001)
  • [5] Kurumatani K., Perspective of multiagent social simulation, ISCIE J Syst Control Inf, 46, pp. 518-523, (2002)
  • [6] Uchibe E., Asada M., Hosoda K., State space construction for cooperative behavior acquisition in the environments including multiple learning robots, J Robot Soc Japan, 20, pp. 281-289, (2002)
  • [7] Edazawa K., Takahashi Y., Asada M., Behavior acquisition in a multi-agent environment using multiple learning modules, 20th Annual Conference of the Robotics Society of Japan, (2002)
  • [8] Goto R., Matsui T., Matsuo H., State generalization with support vector machines in reinforcement learning, 4th Asia-Pacific Conference on Simulated Evolution and Learning, 1, pp. 51-55, (2002)
  • [9] Takahashi Y., Asada M., State-action space construction for multi-layered learning system, J Robot Soc Japan, 21, pp. 164-171, (2003)
  • [10] Gultekin I., Arslan A., Modular-fuzzy cooperation algorithm for multi-agent systems, Advances in Information Systems: 2nd International Conference, pp. 255-263, (2002)
12