AI Seminar

Strategic Reasoning about Network Formation Scenarios

Michael Wellman

Most formal models of network formation posit simple reflexive or myopic behavior on the part of nodes forming connections or selecting local states. Accounting for more sophisticated or forward-looking behavior typically results in a game complicated to a degree precluding tractable analysis, absent heroic simplifying assumptions. We tackle complex network-formation scenarios with an empirical computational approach, combining simulation with game-theoretic reasoning to evaluate hypotheses about the consequences of strategic behavior. Empirical game-theoretic analyses of credit network formation and information cascades illustrate the approach.

Sponsored by