BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Wedges are all you need: sparser and sparser tensor completion
DTSTART:20260402T120000Z
DTEND:20260402T130000Z
DTSTAMP:20260411T150800Z
UID:indico-event-15365@indico.math.cnrs.fr
DESCRIPTION:Speakers: Ludovic Stephan (ENSAI)\n\nTensor completion\, the h
 igher-order analog of matrix completion\, is characterized by a statistica
 l-to-computational gap in the number of samples necessary to retrieve info
 rmation. We show that this is only a consequence of the random sampling sc
 heme: a small modification suffices to bridge the gap between polynomial a
 nd non-polynomial algorithms. We also prove that\, akin to many inference 
 problems\, finding a non-trivial alignment with the signal is the hardest 
 problem\, while the refinement sample complexity does not depend on the te
 nsor order.\n\nhttps://indico.math.cnrs.fr/event/15365/
LOCATION:Fokko du Cloux (ICJ)
URL:https://indico.math.cnrs.fr/event/15365/
END:VEVENT
END:VCALENDAR