This discussion group is intended for (but not limited to) computer scientists doing research in a game theory-related area including (but not limited to) mechanism design, equilibrium finding, e-commerce, etc. The group meets once a week where someone gives a 1-1.5 hour presentation on current research. The emphasis is on research that is not yet published. Presentations are informal (i.e. no slides).

Date | Location | Speaker | Title and abstract |
---|---|---|---|

January 28, 2004 | Wean 4623 | Vincent Conitzer |
A new definition of strategy eliminability spanning a spectrum from
dominant strategies to Nash equilibrium
We like to have a stronger argument than Nash equilibrium for choosing a strategy. But dominant strategies (even iterated dominance) is usually too strong and cannot eliminate enough strategies. I'll discuss a definition of strategy eliminability that spans a spectrum of elimination strength from strict dominance to Nash equilibrium (depending on which sets of strategies you are eliminating relative to). |

February 4, 2004 | No meeting | ||

February 11, 2004 | Wean 4623 | Martin Zinkevich |
My talk today will be on justifications in repeated bimatrix games. A
justification is a set of assumptions about the world and a goal. For
instance, a Bayesian justification consists of a distribution with the
goal of maximizing utility given that belief. A regret justification
consists of an empty set of assumptions and a goal of minimizing
regret. The reason for considering justifications instead of beliefs
is that justifications can be symmetric: meaning, my ASSUMPTION is
that the other agent has the same justification, although it might not
have the same algorithm.
With this basic premise, I will discuss a symmetric justification I have been developing with the goal that it converges to one of a set of Nash equilibria. |

February 18, 2004 | Wean 4623 | Felix Brandt |
(Im)Possibility of Unconditional Privacy in Auctions and Voting
I will investigate the availability of distributed protocols that enable a group of agents to jointly determine the outcome of prominent voting and auction mechanisms by only exchanging messages according to some predetermined protocol rules and without revealing unnecessary information. In particular, privacy should rely neither on trusted third parties (or on a certain fraction of the agents being trusted), nor on computational intractability assumptions (such as the hardness of factoring). |

March 3, 2004 | Wean 4623 | Dan Li | Problems at the intersection of finance and game theory. |

March 10, 2004 | No meeting - Spring Break | ||

March 17, 2004 | Wean 4623 | ???? | ???? |

March 24, 2004 | Wean 4623 | ???? | ???? |

March 31, 2004 | Wean 4623 | ???? | ???? |

April 7, 2004 | Wean 4623 | ???? | ???? |

April 14, 2004 | Wean 4623 | ???? | ???? |

April 21, 2004 | Wean 4623 | ???? | ???? |

April 28, 2004 | Wean 4623 | ???? | ???? |

Last updated: March 2, 2004