Abstract
We investigate how the efficiency of a system performing a density classification task is affected by a modular structure of the network, i.e. a structure where the units can divided in communities. We find that that noise plays a fundamental role in allowing the system to reach global consensus. We also observe that to reach consensus the system needs at least a minimum fraction of the connections to be established outside the communities, and that this fraction depends mainly on two factors: the number of connections per unit and the intensity of noise.
Users
Please
log in to take part in the discussion (add own reviews or comments).