BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:Europe/Vienna
BEGIN:DAYLIGHT
DTSTART:20210328T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20211031T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260406T041111Z
UID:1633449300@ist.ac.at
DTSTART:20211005T175500
DTEND:20211005T185500
DESCRIPTION:Speaker: Dominik Schröder\nhosted by M. Beiglböck\, N. Berest
 ycki\, L. Erdös\, J. Maas\, F. Toninelli\nAbstract: In the first part of 
 the talk we give an overview of recent mathematical results on scaling lim
 its of neural networks with random weights. In particular we introduce the
  neural tangent kernel (NTK) and the conjugate kernel (CK) and explain con
 nections to the double descent phenomenon observed in generalisation perfo
 rmance. In the second part of the talk we report on recent work [Piccolo\,
  S. NeurIPS 2021] on the asymptotic spectrum of the CK in the linear width
  scaling regime. Our result indicates that in the case of an additive bias
  it is impossible to choose an activation function leaving the asymptotic 
 CK spectrum invariant throughout multiple layers\, in sharp contrast to th
 e bias-free case where a simple integral constraint is sufficient for this
  type of invariance.
LOCATION:Heinzel Seminar Room (I21.EG.101)\, Office Building West\, ISTA
ORGANIZER:birgit.oosthuizen-noczil@ist.ac.at
SUMMARY:Dominik Schröder: Neural networks: Evaluation of kernel spectra th
 rough random matrix theory
URL:https://talks-calendar.ista.ac.at/events/3315
END:VEVENT
END:VCALENDAR