Skip to main content

Seminar By Prof Mokshay Madiman

Prof Mokshay Madiman, Yale University, will be giving a IEOR Seminar on this friday, 27th August at 16.15PM.

Topic:   Fundamental Limits of Distributed Estimation and Multiuser Information-Theoretic Games

Abstract: The first part of the talk focuses on fundamental limits of distributed estimation, motivated by a toy model for sensor networks. In distributed estimation, it is of interest to relate the minimax risks of estimating a parameter for users who have
access to different sets of observations. We present some insights into this question in the case of a location parameter, where each user sees either the concatenation or the sum of observations from a set of sources. Surprisingly, the results are closely related to results of fundamental interpretive importance in probability, such as the fact that the entropy of the normalized sums in the central limit theorem increases monotonically to the Gaussian entropy. This part of the talk is joint work with Andrew Barron (Yale), and Abram Kagan and Tinghui Yu (University of Maryland). The second part of the talk reviews basic notions from cooperative game
theory, and exposes a fundamental duality between properties of multiuser scenarios (interpreted as games) on the one hand and
information inequalities on the other. In particular, we outline how this duality can be used to understand certain well known facts about Slepian-Wolf data compression and multiple access channels. Coming full circle, we apply the game theoretic framework to the first part of the talk, and solve a resource allocation problem for our toy sensor network model.

Time and Date:  16.15PM, 27th August, 2010

Venue: Seminar Hall (Room 217), Mech. Engg. Building

News Category