Einladung zum Vortrag im Kolloquium Technische Kybernetik

  Learning in Complex Social Networks

Prof. Munther A. Dahleh
Laboratory for Information and Decision Systems
Massachusetts Institute of Technology
Cambridge · MA
USA

    Zeit: Donnerstag, 27. Mai 2010 · 14:00 Uhr
Ort: IST-Seminarraum 3.243 · Pfaffenwaldring 9 · Campus Stuttgart-Vaihingen

Abstract

We study the (perfect Bayesian) equilibrium of a model of learning over a general social network. Each individual receives a signal about the underlying state of the world, observes the past actions of a stochastically-generated neighborhood of individuals, and chooses one of two possible actions. The stochastic process generating the neighborhoods defines the network topology (social network). We characterize pure-strategy equilibria for arbitrary stochastic and deterministic social networks and characterize the conditions under which there will be asymptotic learning—convergence (in probability) to the right action as the social network becomes large. We show that when private beliefs are unbounded (meaning that the implied likelihood ratios are unbounded), there will be asymptotic learning as long as there is some minimal amount of “expansion in observations”. We present examples showing that "herding" can occur in the absence of such conditions. We also characterize conditions under which there will be asymptotic learning when private beliefs are bounded. Finally, we discuss various robustness issues pertaining to this model.

Biographical Information

Munther A. Dahleh was born in 1962. He received the B.S. degree from Texas A & M university, College Station, Texas in 1983, and his Ph.D. degree from Rice University, Houston, TX, in 1987, all in Electrical Engineering. Since then, he has been with the Department of Electrical Engineering and Computer Science, MIT, Cambridge, MA, where he is now a full Professor. He is currently the associate director of the Laboratory for Information and Decision Systems. He has been a visiting Professor at the Department of Electrical Engineering, California Institute of Technology, Pasadena, CA, for the Spring of 1993. He has held consulting positions with several companies in the US and abroad. Dr. Dahleh has been the recipient of various best paper awards. He became a fellow of IEEE in year 2000. Dr. Dahleh is interested in problems at the interface of robust control, filtering, information theory, and computation which include control problems with communication constraints and distributed mobile agents with local decision capabilities. He is also interested in various problems in network science including distributed computation over noisy network as well as information propagation over complex engineering and social networks. He is also interested in model reduction problems for discrete-alphabet hidden Markov models and universal learning approaches for systems with both continuous and discrete alphabets. He is also interested in the interface between systems theory and neurobiology, and in particular, in providing an anatomically consistent model of the motor control system.