BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//NextToU//EN
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:7e218a15-efba-42c3-8771-c8d9f3188d39@nexttoyou
DTSTAMP:20261105T154202Z
DTSTART:20261105T153000Z
DTEND:20261105T163000Z
SUMMARY:QBio/PEB Seminar - Dmitry Krotov - Dynamical Mind
LOCATION:Yale Science Building 352\, 260 Whitney Avenue\, New Haven\, CT 0
 6511
DESCRIPTION:Dense Associative Memory: physical systems for novel AI archit
 ectures\n\nDense Associative Memories are recurrent neural networks with 
 fixed-point attractor states that are described by an energy function. In
  contrast to conventional Hopfield Networks\, which were popular in the 1
 980s\, Dense Associative Memories have a very large information storage c
 apacity\, making them appealing tools for many problems in AI. In this ta
 lk\, I will provide an intuitive understanding and mathematical framework
  for this class of models and give examples of problems in AI that can be
  tackled using these new ideas. Specifically\, I will explore the relatio
 nship between Dense Associative Memories and transformers. I will present
  a neural network called the Energy Transformer\, which unifies energy-ba
 sed modeling\, associative memories\, and transformers in a single archit
 ecture. I will demonstrate how Energy Transformers can be used for challe
 nging tasks in image processing\, solve partial differential equations\, 
 and serve as computational modules for energy-based language modeling. I 
 will also discuss an exciting possibility of mapping these models onto an
 alog hardware accelerators\, which could enable much more energy-efficien
 t inference compared with GPUs.\n\nHosts: Ben Machta and Michael Abbott\n
 \nhttp://next-to-u.app/events/7e218a15-efba-42c3-8771-c8d9f3188d39
URL:http://next-to-u.app/events/7e218a15-efba-42c3-8771-c8d9f3188d39
END:VEVENT
END:VCALENDAR