BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Stochastic variants of classical optimization methods\, with compl
 exity guarantees - Professor Coralia Cartis
DTSTART:20190501T130000Z
DTEND:20190501T140000Z
UID:TALK114850@talks.cam.ac.uk
CONTACT:J.W.Stevens
DESCRIPTION:Optimization is a key component of machine learning applicatio
 n\, as it helps with training of (neural net\, nonconvex) models and param
 eter tuning. Classical optimization methods are challenged by the scale of
  machine learning applications and the lack of /cost of full derivatives\,
  as well as the stochastic nature of the problem. On the other hand\, the 
 simple approaches that the machine learning community uses need improvemen
 t. Here we try to merge the two perspectives and adapt the strength of cla
 ssical optimization techniques to meet the challenges of data science appl
 ications: from deterministic to stochastic problems\, from typical to larg
 e scale. We propose a general algorithmic framework and complexity analysi
 s that allows the use of inexact\, stochastic and even possibly biased\, p
 roblem information in classical methods for nonconvex optimization. This w
 ork is joint with Katya Scheinberg (Cornell)\, Jose Blanchet (Columbia) an
 d Matt Menickelly (Argonne).
LOCATION:CMS\, MR14
END:VEVENT
END:VCALENDAR