BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260915T085147EDT-0603fD4Vom@132.216.98.100
DTSTAMP:20260915T125147Z
DESCRIPTION:Title:\n\nFeature Learning in Two-layer Neural Networks: The Ef
 fect of Data Covariance\n\nAbstract: \n\nWe study the effect of gradient-b
 ased optimization on feature learning in two-layer neural networks. We con
 sider a setting where the number of samples is of the same order as the in
 put dimension and show that\, when the input data is isotropic\, gradient 
 descent always improves upon the initial random features model in terms of
  prediction risk\, for a certain class of targets. Further leveraging the 
 practical observation that data often contains additional structure\, i.e.
 \, the input covariance has non-trivial alignment with the target\, we pro
 ve that the class of learnable targets can be significantly extended\, dem
 onstrating a clear separation between kernel methods and two-layer neural 
 networks in this regime.\n\n \n
DTSTART:20231002T200000Z
DTEND:20231002T210000Z
LOCATION:Room 1104\, Burnside Hall\, CA\, QC\, Montreal\, H3A 0B9\, 805 rue
  Sherbrooke Ouest
SUMMARY:Murat A. Erdogdu (University of Toronto)
URL:https://www.mcgill.ca/mathstat/channels/event/murat-erdogdu-university-
 toronto-351399
END:VEVENT
END:VCALENDAR