(The talk will be given in English)
Speaker: Prof. Or Ordentlich
School of Computer Science and Engineering, Hebrew University
|
011 hall, Electrical Engineering-Kitot Building
|
Monday, June 17th, 2024
12:00 - 13:00
Strong Data-Processing Inequalities: New Results and Applications
Abstract
The data processing inequality states that mutual information cannot increase along a Markov chain. The strong data processing inequality (SDPI) coefficient gives further information about the rate at which the mutual information decreases over a given channel. When the input distribution is known, even stronger bounds can be obtained using the input-dependent SDPI coefficient. In this talk I will give a short overview on why SDPIs are important and useful, and then present several new results. I will show that the SDPI coefficient of a channel is determined by its "cleanest" binary sub-channel; I will show how to use SDPIs for *lower* bounding the mutual information of a linear code transmitted over a binary channel, as well as the mutual information for n repetitions of the input over a channel; Finally, I will present upper and lower bounds for the input-dependent SDPI coefficient of the Gaussian channel, in terms of the Poincare and log-Sobolev coefficients of the input.
Based on joint works with Uri Erez, Bo’az Klartag, Yury Polyanskiy and Shlomo Shamai (Shitz).
Short Bio
Or Ordentlich received the B.Sc. degree (cum laude) in 2010, M.Sc. degree (summa cum laude) in 2011, and his PhD degree in 2016, all in electrical engineering in Tel Aviv University, Israel. In 2015-2017, he was a postdoctoral fellow in the Laboratory for Information and Decision Systems at the Massachusetts Institute of Technology (MIT), Cambridge and in Boston University. In 2017 he joined the School of Computer Science and Engineering at the Hebrew University of Jerusalem, where he is currently an associate professor. He is also an algorithms architect at Weka. Since 2021 he has been serving as an associate editor for the IEEE Transactions on Information Theory.
השתתפות בסמינר תיתן קרדיט שמיעה = עפ"י רישום שם מלא + מספר ת.ז. בטופס הנוכחות שיועבר באולם במהלך הסמינר
.png)
