Distributed Bayesian learning in multiagent systems: Improving our understanding of its capabilities and limitations

Abstract

We study social networks of agents, where agents learn not only from private signals (i.e., signals only available to the agents receiving them), but from other agents too. Based on all the available information, agents modify their beliefs in events of interest and make decisions on which actions to take based on the beliefs. In doing so, they optimize functions that reflect some (cumulative) reward. This problem has been studied in various disciplines including control theory, operations research, artificial intelligence, game theory, information theory, economics, statistics, computer science, and signal processing.