Some models of information aggregation and consensus in networks
Seminar Room 1, Newton Institute
A primary function of many engineered and social networks is to aggregate the information obtained by the nodes of a network. We discuss a few of the models and thrusts that have been studied, starting with a model of "social learning" by rational (Bayesian) agents, and its connections with information fusion models in the engineering literature. We then consider a set of agents (processors, decision makers, sensors, etc.) who reach consensus through an iterative process involving the exchange and averaging of local values. We discuss a number of models, application contexts, and convergence results, and the connections with Markov chain theory.