BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium: Understanding machine learning via exactly solvable
  models
DTSTART:20200217T101500
DTEND:20200217T111500
DTSTAMP:20260407T110927Z
UID:3446d4418bc84526f3c6f87a06f06c22f8d82a5ddbc2e59eae2b86be
CATEGORIES:Conferences - Seminars
DESCRIPTION:By: Lenka Zdeborova\, Institute of Theoretical Physics CEA Sac
 lay\nIC Faculty candidate\n\nAbstract:\nThe affinity between statistical p
 hysics and machine learning has long history. I will describe the main lin
 es of this long-lasting friendship in the context of current theoretical c
 hallenges and open questions about deep learning. Theoretical physics ofte
 n proceeds in terms of solvable synthetic models\, I will describe the rel
 ated line of work on solvable models of simple feed-forward neural network
 s. I will highlight a path forward to capture the subtle interplay between
  the structure of the data\, the architecture of the network\, and the opt
 imization algorithms commonly used for learning.  \n\nBio:\nLenka Zdebor
 ová is a researcher at CNRS working in the Institute of Theoretical Physi
 cs in CEA Saclay\, France. She received a PhD in physics from University P
 aris-Sud and from Charles University in Prague in 2008. She spent two year
 s in the Los Alamos National Laboratory as the Director's Postdoctoral Fel
 low. In 2014\, she was awarded the CNRS bronze medal\, in 2016 Philippe Me
 yer prize in theoretical physics and an ERC Starting Grant\, in 2018 the I
 rène Joliot-Curie prize. She is editorial board member for Journal of Phy
 sics A\, Physical review E\, Physical Review X and SIMODS. Lenka's experti
 se is in applications of methods developed in statistical physics\, such a
 s advanced mean field methods\, replica method and related message passing
  algorithms\, to problems in machine learning\, signal processing\, statis
 tical inference and optimization. \n\nMore information
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR