BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Quest | CBMM Seminar Series
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20261012T002710Z
UID:tag:localist.com\,2008:EventInstance_47305452408342
DTSTART:20250211T210000Z
DTEND:20250211T220000Z
DESCRIPTION:Dr. Thomas Serre\, Thomas J. Watson\, Sr. Professor of Science 
 at Brown University's Carney Center for Computational Brain Science\, to s
 peak as part of the Quest for Intelligence and CBMM seminar series.\n\n \n
 \nTitle: Aligning deep networks with human vision will require novel neura
 l architectures\, data diets and training algorithms\n\nAbstract: Recent a
 dvances in artificial intelligence have been mainly driven by the rapid sc
 aling of deep neural networks (DNNs)\, which now contain unprecedented num
 bers of learnable parameters and are trained on massive datasets\, coverin
 g large portions of the internet. This scaling has enabled DNNs to develop
  visual competencies that approach human levels. However\, even the most s
 ophisticated DNNs still exhibit strange\, inscrutable failures that diverg
 e markedly from human-like behavior—a misalignment that seems to worsen 
 as models grow in scale.\n\nIn this talk\, I will discuss recent work from
  our group addressing this misalignment via the development of DNNs that m
 imic human perception by incorporating computational\, algorithmic\, and r
 epresentational principles fundamental to natural intelligence. First\, I 
 will review our ongoing efforts in characterizing human visual strategies 
 in image categorization tasks and contrasting these strategies with modern
  deep nets. I will present initial results suggesting we must explore nove
 l data regimens and training algorithms for deep nets to learn more human-
 like visual representations. Second\, I will show results suggesting that 
 neural architectures inspired by cortex-like recurrent neural circuits off
 er a compelling alternative to the prevailing transformers\, particularly 
 for tasks requiring visual reasoning beyond simple categorization.
GEO:42.362302;-71.091766
LOCATION:Building 46\, Singleton Auditorium 46-3002
SUMMARY:Quest | CBMM Seminar Series
URL;VALUE=URI:https://calendar.mit.edu/event/quest-cbmm-seminar-series-797
CATEGORIES:Conferences/Seminars/Lectures
END:VEVENT
END:VCALENDAR