BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:The role of Random  Matrix Theory for the optimization algorithms 
 of Machine Learning  - Gérard Ben Arous  (Courant)
DTSTART:20260312T163000Z
DTEND:20260312T173000Z
UID:TALK234784@talks.cam.ac.uk
CONTACT:Richard Samworth
DESCRIPTION:I will survey recent progress  in the understanding of the opt
 imization dynamics for important tasks for Machine Learning or high dimens
 ional statistics. We will see how these very high-dimensional dynamics are
  in fact ruled by the so-called "effective dynamics" of much lower dimensi
 onal systems. This dynamical dimension reduction is related to the BBP spe
 ctral transition of Random Matrix Theory\,  appearing dynamically along th
 e algorithm path. I will illustrate these phenomena in multi-spike Tensor 
 PCA\, XOR\, and classification of Gaussian mixtures with multi-layer neura
 l nets. \n\nThis talk is based on joint works with Reza Gheissari (Northwe
 stern)\, Jiaoyang Huang (Wharton)\, Aukosh Jagannath (Waterloo)\, and on j
 oint works with Cedric Gerbelot (ENS Lyon) and Vanessa Piccolo (EPFL).\n\n
 A wine reception in the Central Core will follow the lecture. 
LOCATION:Centre for Mathematical Sciences MR2
END:VEVENT
END:VCALENDAR