A new Q-learning algorithm based on the Metropolis criterion

Summary, in English

The balance between exploration and exploitation is one of the key problems of action selection in Q-learning. Pure exploitation causes the agent to reach the locally optimal policies quickly, whereas excessive exploration degrades the performance of the Q-learning algorithm even if it may accelerate the learning process and allow avoiding the locally optimal policies. In this paper, finding the optimum policy in Q-learning is de scribed as search for the optimum solution in combinatorial optimization. The Metropolis criterion of simulated annealing algorithm is introduced in order to balance exploration and exploitation of Q-learning, and the modified Q-learning algorithm based on this criterion, SA-Q-learning, is presented. Experiments show that SA-Q-learning converges more quickly than Q-learning or Boltzmann exploration, and that the search does not suffer of performance degradation due to excessive exploration.

• Author: MZ Guo; Y Liu; Jacek Malec
• Department/s: Computer Science
• Publishing year: 2004
• Language: English
• Pages: 2140-2143
• Publication/Series: IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
• Volume: 34
• Issue: 5
• Links: http://dx.doi.org/10.1109/TSMCB.2004.832154
• Document type: Journal article
• Publisher: IEEE - Institute of Electrical and Electronics Engineers Inc.
• Topic: Computer Science
• Keywords: reinforcement learning; Q-learning; metropolis criterion; exploitation; exploration
• Status: Published
• ISBN/ISSN/Other: ISSN: 1083-4419