BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.4.6//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:960@fds.yale.edu
DTSTART;TZID=America/New_York:20261014T120000
DTEND;TZID=America/New_York:20261014T130000
DTSTAMP:20261012T155631Z
URL:https://fds.yale.edu/events/fds-colloquium-boris-hanin/
SUMMARY:FDS Colloquium: Boris Hanin (Princeton)\, "Neural Network Scaling L
 imits: From Theory to Practice"
DESCRIPTION:\n\n\nAbstract: A fundamental question in deep learning is to d
 escribe the scaling limits of neural networks by answering how and what ne
 ural networks learn in regimes where both model and dataset size diverge. 
 From the perspective of theory the goal is to identify what infinite size 
 limits of neural network training dynamics exist and to analyze the mechan
 isms by which architecture and optimizer combine to learn data-dependent f
 eatures. The aspiration is to develop an analog of the ideal gas law\, whe
 re out of the collective motion of many weakly interacting molecules (in t
 his analogy network parameters) emerge a small number of thermodynamic var
 iables like pressure\, temperature\, and volume that give simple and predi
 ctive insights into the properties of realistic gasses. From the practical
  perspective the objective is to design a wind tunnel for neural networks.
  The goal is to predict reliably from experiments on small models and data
 sets the most performant ways to align architecture\, data\, and optimizer
  at large scale. In this talk I will give a survey of some of my work on n
 eural network scaling limits emphasizing the interplay between theory and 
 practice.\n\n\n\nSpeaker Bio: I am an Associate Professor at Princeton OR
 FE (and Associated Faculty at Princeton PACM) working on deep learning\
 , probability\, and spectral asymptotics. Prior to Princeton\, I was an A
 ssistant Professor in Mathematics at Texas A&amp\;M\, an NSF Postdoc at MI
 T Math\, and a PhD student in Math at Northwestern\, where I was supervise
 d by Steve Zelditch.\n\n\n\nI am one of the co-organizers of the&nbsp\;Pri
 nceton Probability Seminar&nbsp\;and the&nbsp\;alg-ML seminar.\n\n\n\nI am
  an advisor at&nbsp\;Mithril\,&nbsp\;Logical Intelligence\,&nbsp\;Lumos He
 alth\, and&nbsp\;9D Labs.\n\n\n\n\n
CATEGORIES: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:DAYLIGHT
DTSTART:20260308T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR