BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260715T162913EDT-8694CjG9X9@132.216.98.100
DTSTAMP:20260715T202913Z
DESCRIPTION:\n	Approximate Cross-Validation for Large Data and High Dimensio
 ns\n\n	 \n\n	Abstract:\n\n\nThe error or variability of statistical and mach
 ine learning algorithms is often assessed by repeatedly re-fitting a model
  with different weighted versions of the observed data. The ubiquitous too
 ls of cross-validation (CV) and the bootstrap are examples of this techniq
 ue. These methods are powerful in large part due to their model agnosticis
 m but can be slow to run on modern\, large data sets due to the need to re
 peatedly re-fit the model. We use a linear approximation to the dependence
  of the fitting procedure on the weights\, producing results that can be f
 aster than repeated re-fitting by orders of magnitude. This linear approxi
 mation is sometimes known as the “infinitesimal jackknife” (IJ) in the sta
 tistics literature\, where it has mostly been used as a theoretical tool t
 o prove asymptotic results. We provide explicit finite-sample error bounds
  for the infinitesimal jackknife in terms of a small number of simple\, ve
 rifiable assumptions. Without further modification\, though\, we note that
  the IJ deteriorates in accuracy in high dimensions and incurs a running t
 ime roughly cubic in dimension. We additionally show\, then\, how dimensio
 nality reduction can be used to successfully run the IJ in high dimensions
  when data is sparse or low rank. Simulated and real-data experiments supp
 ort our theory.\n\n\n	Speaker\n\n\nTamara Broderick is an Associate Profess
 or in the Department of Electrical Engineering and Computer Science at MIT
 . She is a member of the MIT Computer Science and Artificial Intelligence 
 Laboratory (CSAIL)\, the MIT Statistics and Data Science Center\, and the 
 Institute for Data\, Systems\, and Society (IDSS). She completed her Ph.D.
  in Statistics at the University of California\, Berkeley in 2014. Previou
 sly\, she received an AB in Mathematics from Princeton University (2007)\,
  a Master of Advanced Study for completion of Part III of the Mathematical
  Tripos from the University of Cambridge (2008)\, an MPhil by research in 
 Physics from the University of Cambridge (2009)\, and an MS in Computer Sc
 ience from the University of California\, Berkeley (2013). Her recent rese
 arch has focused on developing and analyzing models for scalable Bayesian 
 machine learning. She has been awarded an Early Career Grant (ECG) from th
 e Office of Naval Research (2020)\, an AISTATS Notable Paper Award (2019)\
 , an NSF CAREER Award (2018)\, a Sloan Research Fellowship (2018)\, an Arm
 y Research Office Young Investigator Program (YIP) award (2017)\, Google F
 aculty Research Awards\, an Amazon Research Award\, the ISBA Lifetime Memb
 ers Junior Researcher Award\, the Savage Award (for an outstanding doctora
 l dissertation in Bayesian theory and methods)\, the Evelyn Fix Memorial M
 edal and Citation (for the Ph.D. student on the Berkeley campus showing th
 e greatest promise in statistical research)\, the Berkeley Fellowship\, an
  NSF Graduate Research Fellowship\, a Marshall Scholarship\, and the Phi B
 eta Kappa Prize (for the graduating Princeton senior with the highest acad
 emic average).\n\nZoom Link\n\n \n
DTSTART:20201113T203000Z
DTEND:20201113T213000Z
SUMMARY:Tamara Broderick (MIT)
URL:https://www.mcgill.ca/mathstat/channels/event/tamara-broderick-mit-3262
 04
END:VEVENT
END:VCALENDAR