Calendrier du 09 octobre 2017
Roy Seminar (ADRES)
Du 09/10/2017 de 17:00 à 18:30
Salle R1-09, Campus Jourdan, 48 boulevard Jourdan, 75014 Paris
WOLITZKY Alexander (MIT)
Learning from Others' Outcomes
Abstract: I develop a simple model of social learning in which players observe others' outcomes but not their actions. There is a continuum of players, and each player chooses once-and-for-all between a safe action (which succeeds with known probability) and a risky action (which succeeds with fixed but unknown probability, depending on the state of the world). The actions also differ in their costs. Before choosing, a player observes the outcomes of K others. There is always an equilibrium in which success is more likely in the good state, and this regularity property holds whenever the initial generation of players is not well-informed about the state. In the case of an outcome-improving innovation (where the risky action may yield a higher probability of success), players take the correct action as K??. In the case of a cost-saving innovation (where the risky action involves saving a cost but accepting a lower probability of success), inefficiency persists as K?? in any regular equilibrium. Whether inefficiency takes the form of under-adoption or over-adoption also depends on the nature of the innovation. Convergence of the population to equilibrium may be non-monotone.
Paris Game Theory Seminar
Du 09/10/2017 de 11:00 à 12:00
Salle 01 (RDC), Centre Emile Borel, Institut Henri Poincaré, 11 rue Pierre et Marie Curie, 75005 Paris
MERTIKOPOULOS Panayotis ()
No-Regret Learning in Games
In many cases of practical interest, the players of a repeated game may not know the structure of the game being played - simply think of commuters driving to work every day, ignorant of the number of commuters at each part of the road. In such cases, it is often assumed that players follow a no-regret procedure, i.e. an updating policiy that provably minimizes each player's individual regret against any possible play of their opponents.
This talk focuses on the following question: does the sequence of play induced by a no-regret learning process converge to an equilibrium of the underlying stage game? I will present some recent contributions to this question (in both finite and continuous games), and I will discuss the impact of the feedback available to the players.