Network Models and Algorithms for Game Theory and Economics (Research Seminar, October 7th, 2004) Michael Kearns University of Pennsylvania Abstract In this talk I will survey work we have done over the last few years on graph-theoretic models for strategic interaction in large-population games and economic systems. In such models, the presence of an edge between two parties indicates direct strategic influence (via payoffs or exchange), in much the same way as the edges of graphical models for probabilistic inference (such as Bayes and Markov nets) indicate direct stochastic influence. Our particular interests in such models include: Their representational and computational benefits for equilibrium and other computations Their relationship with graphical models for probabilistic inference The relationships between network structure and equilibrium outcomes * Joint work with Sham Kakade, John Langford, Michael Littman, Luiz Ortiz, Robin Pemantle, Satinder Singh and Siddarth Suri. Related papers: 1   2   3   4   5