Information theory data science and the curse of phase transitions

by Marc Mezard

at Physics Colloquium

Tue, 13 Jun 2017, 15:00
Nanotechnology institute building (#51) room 15

Abstract

Statistical physics is dedicated to understanding the collective behavior of systems with many interacting particles In the last forty years its major breakthroughs in dealing with strongly disordered systems glasses have opened the way to using its concepts and methods in other disciplines opening what Philip Anderson once called the spin glass cornucopia This talk will show how key concepts of statistical physics phase transitions and the occurrence of glassy phases play a major role in some core problems of information theory and data analysis This analysis will be illustrated by specific examples in data transmission error correcting codes and compressed data acquisition compressed sensing Refreshments are served at 14:45

Created on 05-06-2017 by Bar Lev, Yevgeny (ybarlev)
Updaded on 05-06-2017 by Bar Lev, Yevgeny (ybarlev)