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:20260405T191729Z
UID:1626357600@ist.ac.at
DTSTART:20210715T160000
DTEND:20210715T170000
DESCRIPTION:Speaker: Alexander Kolesnikov\nhosted by Marco Mondelli\nAbstra
 ct: In this talk I will demonstrate that transfer of visual representation
 s pre-trained on large-scale data can dramatically improve sample efficie
 ncy and simplify hyperparameter tuning. In the first part of the talk I w
 ill discuss challenges that arise in large-scale pre-training and how to a
 ddress them. Then I will dive into strategies for adapting pre-trained mod
 els for a target task. Finally\, I will present extensive empirical evalua
 tion of large-scale visual models and highlight many surprising findings.
  In particular\, it turns out that huge models pre-trained on large data n
 ot only achieve state-of-the-art performance on many standard vision bench
 marks\, but are also very strong few-shot learners and generalize well in 
 out-of-distribution evaluation scenarios.
LOCATION:Zoom Link: Join Zoom Meeting https://istaustria.zoom.us/j/91703991
 543?pwd=dTl3VWlWVTl1MGxxRjBKakN0ajRPZz09  Meeting ID: 917 0399 1543 Passco
 de: 378221\, ISTA
ORGANIZER:
SUMMARY:Alexander Kolesnikov: Learning general visual representation with l
 arge-scale pre-training
URL:https://talks-calendar.ista.ac.at/events/3235
END:VEVENT
END:VCALENDAR