BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20261113T154912EST-0337Ino7AR@132.216.98.100
DTSTAMP:20261113T204912Z
DESCRIPTION:Title:\n\nSharply predicting the behavior of iterative algorith
 ms in random nonconvex optimization problems\n\nAbstract: \n\nIterative al
 gorithms are the workhorses of modern statistical learning\, and are widel
 y used to fit large-scale\, complex models to random data. While the choic
 e of an algorithm and its hyperparameters determines both the speed and fi
 delity of the learning pipeline\, it is common for this choice to be made 
 heuristically\, either by expensive trial-and-error or by comparing upper 
 bounds on convergence rates of various candidate algorithms. Motivated by 
 this\, we develop a principled framework that produces sharp\, iterate-by-
 iterate characterizations of solution quality for a wide variety of algori
 thms on several nonconvex model-fitting problems with random data. I will 
 present the general framework and then some concrete consequences\, showca
 sing how sharp predictions can provide precise separations between familie
 s of algorithms while also revealing some nonstandard convergence phenomen
 a. The talk will be based on joint work with Kabir Chandrasekher\, Mengqi 
 Lou\, and Christos Thrampoulidis.\n
DTSTART:20230320T203000Z
DTEND:20230320T213000Z
LOCATION:Room 1104\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue
  Sherbrooke Ouest
SUMMARY:Ashwin Pananjady (Georgia Tech)
URL:https://www.mcgill.ca/mathstat/channels/event/ashwin-pananjady-georgia-
 tech-346821
END:VEVENT
END:VCALENDAR