BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:922@fds.yale.edu
DTSTART;TZID=America/New_York:20260205T160000
DTEND;TZID=America/New_York:20260205T170000
DTSTAMP:20260120T200614Z
URL:https://fds.yale.edu/events/fds-colloquium-yuhai-tu-flatiron-physics-f
 or-deep-learning-towards-a-theoretical-foundation/
SUMMARY:FDS Colloquium: Yuhai Tu (Flatiron)\, "Physics for Deep Learning: T
 owards a Theoretical Foundation"
DESCRIPTION:\nTalk summary: Artificial Neural Network (ANN) and Machine Lea
 rning (ML) have received a huge amount of attention among physicists trigg
 ered by the 2024 Nobel Physics Prize to John Hopfield and Geoff Hinton. Hi
 storically\, ANN-based ML was developed based on ideas from neuroscience a
 nd statistical physics. However\, the recent successes of deep learning ne
 ural networks (DLNN) are mainly driven by large amount of data and exponen
 tially growing computing capability.  As a result\, despite its successes
  in many different disciplines\, DLNN remains largely a black box – it i
 s unclear how it learns and whether what it learns is generalizable.  In 
 the past several years\, we have been trying to develop a theoretical fram
 ework based on statistical physics and stochastic dynamical systems theory
  to study DLNN. In this talk\, we will first give a broad introduction of 
 the ANN-based ML emphasizing on the role of physics in its development fol
 lowed by describing some of our recent progress in understanding DLNN in t
 wo related areas – learning dynamics and generalization.\n\n\n\nSpeaker 
 bio: Yuhai Tu joined the Center for Computational Biology with a joint pos
 ition at the Center for Computational Neuroscience in May 2025 as a Senior
  Research Scientist. Yuhai’s work focuses on Statistical Physics\, Molec
 ular/Cellular Biology\, Neuroscience\, and Machine Learning. His pioneerin
 g work on collective phenomena in active systems (flocking dynamics) in th
 e 90’s won him (together with John Toner and Tamas Vicsek) the APS Lars 
 Onsager Prize in 2020. Since 2000\, his research interests shift to biolog
 ical physics. He has made seminal contributions in many areas of biologica
 l physics including algorithm development and statistical analysis for hig
 h throughput transcriptome data (microarray analysis)\; quantitative model
 ing of signal transduction and motility in bacterial chemotaxis\; and ther
 modynamics of nonequilibrium biochemical networks. His recent work focuses
  on three directions: (1) dynamics of biological networks — biochemical 
 networks for signal transduction and neural networks for coding and comput
 ation\; (2) thermodynamics of information processing in biological systems
 \; (3) statistical physics of machine learning.\n\n\n\nYuhai Tu graduated 
 from the School of Gifted Young at University of Science and Technology of
  China in 1987. He came to the US under the CASPEA program and got his Ph.
 D. in theoretical physics from University of California San Diego (UCSD) i
 n 1991. After 3 years as the Division Prize Fellow at Caltech\, he joined 
 IBM T. J. Watson Research Center from 1994-2025 and served as head of the 
 theory group during 2002-2014. He joined Flatiron Institute in 2025. He is
  an American Physical Society (APS) fellow (elected 2004)\, and an America
 n Association for the Advancement of Science (AAAS) fellow (elected 2020).
  He served as Vice-Chair/Chair of APS Division of Biological Physics (DBIO
 ) during 2016-2018.\n
CATEGORIES:Fellows Events,FDS Events,Colloquium
LOCATION:Yale Institute for Foundations of Data Science\, Kline Tower 13th 
 Floor\, Room 1327\, New Haven\, CT\, 06511\, United States
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=Kline Tower 13th Floor\, Ro
 om 1327\, New Haven\, CT\, 06511\, United States;X-APPLE-RADIUS=100;X-TITL
 E=Yale Institute for Foundations of Data Science:geo:0,0
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:STANDARD
DTSTART:20251102T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
END:STANDARD
END:VTIMEZONE
END:VCALENDAR