Program Overview
| Time | Monday | Tuesday | Wednesday | Thursday | Friday |
|---|---|---|---|---|---|
|
Morning Session I
Ìý(9:00 - 10:30)
|
Frans WillemsÌý | Gerhard Kramer | Michael GastparÌý | Sennur Ulukus | Alex DimakisÌý |
|
ÌýCoffee Break
(10:30 - 11:00)
|
--Ìý | -- | --Ìý | --Ìý | --Ìý |
|
Morning Session II
Ìý(11:00 - 12:30) ÌýÌý
|
Frans WillemsÌýÌý | Gerhard Kramer | Michael Gastpar | Sennur UlukusÌýÌý | Alex DimakisÌýÌý |
| Lunch (12:30 - 14:00 )Ìý |
--Ìý | --Ìý | --Ìý | --Ìý | --Ìý |
|
Afternoon Session I
Ìý(14:00 - 15:30) Ìý |
Student ÌýPresentationsÌý Session IÌý |
StudentÌý Presentations Session III |
ExcursionÌý | ÌýAmos Lapidoth | School ClosesÌý |
|
Coffee Break
(15:30 - 16:00) Ìý |
-- | --Ìý | ExcursionÌý | --Ìý | Ìý |
|
Afternoon Session II
Ìý(16:00 - 18:00) Ìý |
Student Presentations Session IIÌýÌýÌý |
Student Presentations Session IVÌýÌýÌý |
Excursion | Student Presentations ÌýSession VÌýÌýÌýÌýÌý |
Ìý |
| Ìý | Ìý | Ìý | Ìý | Ìý | Ìý |
Ìý
Lectures:Ìý
Ìý
Frans Willems Ìý (Eindhoven University of Technology, Netherlands)
Introduction to Universal Source Coding and Biometrics
Monday, April 16, 9:00 - 12:30
Ìý
The lecture consists of two parts. The first part focusses on universal source coding, the second part on biometrics.ÌýIn the first part we consider universal source coding based on waiting times and coding based on weighting over tree models. The waiting-time algorithm hinges on Kacs theorem [1947]. It achieves entropy for stationary ergodic sources and explains the idea underlying the Lempel-Ziv buffer source coding method [1977].ÌýThe weighting-method is based on context trees and arithmetic coding. It achieves entropy for tree sources but moreover minimizes the redundancy. The algorithm can also be applied for learning the tree model producing the data that are observed.
Ìý
In the second part of the lecture we consider biometrics from an information-theoretical perspective. We will consider both authentication and identification protocols, and investigate protected and unprotected databases. The emphasis is on binary biometrics and error exponents.
Ìý
Gerhard KramerÌý (Technical University of Munich, Germany)
Network Flow and Network Coding
Tuesday, April 17, 9:00 - 12:30
Ìý
The lectures treat the basics of network flow via Ford and Fulkerson's flow-augmenting path algorithm and
of linear network coding via Koetter and Medard's linear systems approach. Time-permitting, we show how
the results generalize to classic networks using random network coding, and to generalÌýnetworks, including
wireless networks, using "noisy" or "digital" network coding by viewing network coding as a compress-forward
relaying strategy.
Ìý
Michael Gastpar (UC Berkeley, USA and EPFL, Switzerland)
AlgebraicÌýStructure in Network Information Theory
Wednesday, April 18, 9:00 - 12:30
Ìý
In information-theoretic theorems, achievability arguments are typically established via the IID random coding argument: codewords are sampled randomly from a judiciously chosen probability distribution. By contrast, algebraic methods are used to develop practical encoding and decoding schemes of reasonable complexit