BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Superpower Daily//AI Events//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Superpower Daily AI events
BEGIN:VEVENT
UID:14898618-31b2-45a5-9b8b-64336eab9289@superpowerdaily.com
DTSTAMP:20261024T140356Z
LAST-MODIFIED:20261024T140356Z
DTSTART:20261029T183000Z
DTEND:20261029T193000Z
SUMMARY:Making Data Count: Measuring\, Stretching\, and Amortizing Training
  Data
DESCRIPTION:Harvard SEAS Assistant Professor David Alvarez-Melis will prese
 nt a data-centric approach to AI\, exploring how optimal transport can mea
 sure datasets\, scaling laws can extend scarce training data\, and boomera
 ng distillation can produce models at intermediate sizes. His talk will al
 so consider open questions for building a stronger foundation for machine 
 learning.\nDetails: https://superpowerdaily.com/events/making-data-count-m
 easuring-stretching-and-amortizing-training-data-14898618-31b2-45a5-9b8b-6
 4336eab9289\nRegister: https://events.seas.harvard.edu/event/computer-scie
 nce-colloquium-series-david-alvarez-melis-harvard-seas
LOCATION:Allston\, MA
URL:https://superpowerdaily.com/events/making-data-count-measuring-stretchi
 ng-and-amortizing-training-data-14898618-31b2-45a5-9b8b-64336eab9289
STATUS:TENTATIVE
TRANSP:TRANSPARENT
END:VEVENT
END:VCALENDAR