Multi-agent learning and the descriptive value of simple models
作者：
Highlights：
•
摘要
Behavioral research suggests that human learning in some multi-agent systems can be predicted with surprisingly simple “foresight-free” models. The current note discusses the implications of this research, and its relationship to the observation that social interactions tend to complicate learning.
论文关键词：Reinforcement learning,Fictitious play,Equivalent Number of Observations (ENO),Reciprocation
论文评审过程：Received 25 April 2006, Revised 20 October 2006, Accepted 22 November 2006, Available online 26 January 2007.
论文官网地址：https://doi.org/10.1016/j.artint.2007.01.001
