BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Journée de Rham 2017 : Mathematical Mysteries of Deep Networks
DTSTART:20170308T153000
DTEND:20170308T163000
DTSTAMP:20260916T045125Z
UID:45811562faf4756aa978acbe734555c1f2c97b30036b8756faead601
CATEGORIES:Conferences - Seminars
DESCRIPTION:Stéphane Mallat\, Ecole Normale Supérieure\, Paris\nDeep con
 volutional networks provide state of the art classifications and regressio
 ns results over many high-dimensional problems including image\, audio and
  bio-medical signal recognition\, with spectacular results. They are able 
 to approximate high-dimensional functionals without suffering from the cur
 se of dimensionality. We shall review their architecture and concentrate o
 n the analysis of their mathematical properties\, which are mostly not und
 erstood. Applications will be shown for image and audio processing\, but a
 lso to approximate complex stochastic processes for statistical physics\, 
 and quantum chemistry.\n 
LOCATION:CO 2 https://plan.epfl.ch/theme/generalite_thm_v2?request_locale=
 fr&room=CO%202&domain=places&dim_floor=1&lang=fr&dim_lang=fr&tree_groups=c
 entres_nevralgiques%2Cacces%2Cmobilite_reduite%2Censeignement%2Cco
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR