Claudia Winklmayr: Collective Decision Making under Correlated Information
BCCN Berlin and TU Berlin
Abstract
Many aspects of collective behavior can be framed as binary decisions between a correct and an incorrect option. For such cases it can be shown that a group of independent decision makers, deciding via majority vote, will often display higher accuracy than any individual group member. This phenomenon has been termed the Wisdom of Crowds. In many real-world scenarios, however, the individual votes will not be statistically independent but rather show some degree of correlation. In this thesis we study the impact of correlated information on collective decisions.
As a foundation we use the two-cue model developed by Kao and Couzin in (Kao and Couzin, 2014), in which a fraction of group members samples information from a shared source. We then extend this model in two ways: firstly to the temporal domain, by framing the individual decision process as an integration of information over time, and secondly to the network domain, by viewing agents as nodes in a Watts-Strogatz network and modelling opinion exchange via (a) the voter model and (b) the threshold model.
Organized by
Pawel Romanczuk / Robert Martin
Location
BCCN Berlin, lecture hall, Philippstr. 13 Haus 6, 10115 Berlin