Loading...

Towards a bounded-rationality model of multi-agent social learning in games

Hemmati, M ; Sharif University of Technology | 2010

783 Viewed
  1. Type of Document: Article
  2. DOI: 10.1109/ISDA.2010.5687277
  3. Publisher: 2010
  4. Abstract:
  5. This paper deals with the problem of multi-agent learning of a population of players, engaged in a repeated normal-form game. Assuming boundedly-rational agents, we propose a model of social learning based on trial and error, called "social reinforcement learning". This extension of well-known Q-learning algorithm, allows players within a population to communicate and share their experiences with each other. To illustrate the effectiveness of the proposed learning algorithm, a number of simulations on the benchmark game of "Battle of Sexes" has been carried out. Results show that supplementing communication to the classical form of Q-learning, significantly improves convergence speed towards Nash equilibrium
  6. Keywords:
  7. Agent-based model ; Convergence speed ; Multi-Agent ; Multi-agent learning ; Nash equilibrium ; Population game ; Q-learning ; Q-learning algorithms ; Rational agents ; Social learning ; Trial and error ; Computational methods ; Intelligent systems ; Reinforcement learning ; Systems analysis ; Learning algorithms
  8. Source: 2010 10th International Conference on Intelligent Systems Design and Applications, ISDA'10, Cairo, 29 November 2010 through 1 December 2010 ; 2010 , Pages 142-148 ; 9781424481354 (ISBN)
  9. URL: http://ieeexplore.ieee.org/document/5687277/?reload=true