BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.4.0.1//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:576@fds.yale.edu
DTSTART;TZID=America/New_York:20230329T160000
DTEND;TZID=America/New_York:20230329T170000
DTSTAMP:20250916T142121Z
URL:https://fds.yale.edu/events/fds-colloquium-nathan-srebro-ttic-interpol
 ation-learning-and-overfitting-with-linear-predictors-and-short-programs/
SUMMARY:FDS Colloquium: Nathan Srebro (TTIC) “Interpolation Learning and
  Overfitting with Linear Predictors and Short Programs”
DESCRIPTION:"Interpolation Learning and Overfitting with Linear Predictors
  and Short Programs"\n\n\nLocation: Mason 211 or remote access: https://ya
 le.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=7e9e0891-7848-44ad-91e7
 -af93011fd580 \n\n\n\n\n\n\n\nSpeaker: Nathan SrebroProfessor\, Toyota Tec
 hnological Institute at Chicago\n\n\n\nAbstract: Classical theory\, conven
 tional wisdom\, and all textbooks\, tell us to avoid reaching zero trainin
 g error and overfitting the noise\, and instead balance model fit and comp
 lexity.  Yet\, recent empirical and theoretical results suggest that in m
 any cases overfitting is benign\, and even interpolating the training data
  can lead to good generalization.  Can we characterize and understand whe
 n overfitting is indeed benign\, and when it is catastrophic as classic t
 heory suggests?  And can existing theoretical approaches be used to stud
 y and explain benign overfitting and the "double descent" curve?  I will 
 discuss interpolation learning in linear (and kernel) methods\, as well as
  using the universal "minimum description length" or "shortest program" le
 arning rule.\n\n\n\n\n\n\n\nBio: Nati (Nathan) Srebro is a professor at t
 he Toyota Technological Institute at Chicago\, with cross-appointments at 
 the University of Chicago's Department of Computer Science\, and Committee
  on Computational and Applied Mathematics. He obtained his PhD from the Ma
 ssachusetts Institute of Technology in 2004\, and previously was a postdoc
 toral fellow at the University of Toronto\, a visiting scientist at IBM\, 
 and an associate professor at the Technion.  \n\n\n\nDr. Srebro’s rese
 arch encompasses methodological\, statistical and computational aspects of
  machine learning\, as well as related problems in optimization. Some of S
 rebro’s significant contributions include work on learning “wider” M
 arkov networks\, introducing the use of the nuclear norm for machine learn
 ing and matrix reconstruction\, work on fast optimization techniques for m
 achine learning\, and on the relationship between learning and optimizatio
 n. His current interests include understanding deep learning through a det
 ailed understanding of optimization\, distributed and federated learning\,
  algorithmic fairness and practical adaptive data analysis.\n
CATEGORIES:FDS Events,Colloquium,Seminar Series
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
DTSTART:20230312T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR