BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.4.0.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:361@isdm.umontpellier.fr
DTSTART;TZID=Europe/Paris:20250401T140000
DTEND;TZID=Europe/Paris:20250401T140000
DTSTAMP:20260828T094259Z
URL:https://isdm.umontpellier.fr/events/end-to-end-private-learning-challe
 nges-of-noisy-and-inconsistent-data-2/
SUMMARY:End-to-End Private Learning: Challenges of Noisy and Inconsistent D
 ata
DESCRIPTION:Online\n\nMachine Learning in Montpellier\, Theory &amp\; Pract
 ice - Shubhankar Mohapatra (University of Waterloo)\n\nMany critical AI ap
 plications\, such as personalized health assistants and social network rec
 ommender systems\, rely on learning from private data. Differential privac
 y has become the gold standard for training models on sensitive data\, gai
 ning widespread adoption in industry and government. This growing adoption
  has fueled research in differentially private learning\, yet significant 
 challenges remain in its deployment. In this talk\, I will discuss several
  challenges that arise when integrating differential privacy into an end-t
 o-end learning pipeline\, from data collection to model training. I will p
 articularly focus on the difficulties caused by inconsistencies in private
  datasets\, such as typos and missing values. Correcting these errors is e
 specially challenging when direct access to the raw data is restricted due
  to privacy constraints. I will present two recent works addressing this i
 ssue. First\, I will show how leveraging correlations and dependencies in 
 the data lets us privately quantify inconsistencies\, helping estimate dat
 a quality and the effort needed for data repair. Second\, I will examine t
 he impact of missing values in private learning and introduce simple yet e
 ffective techniques for generating differentially private synthetic data t
 o mitigate these effects.\n\nMachine Learning in Montpellier\, Theory &amp
 \; Practice
ATTACH;FMTTYPE=image/jpeg:https://isdm.umontpellier.fr/wp-content/uploads/
 2026/06/ml-mtp-gC78d5.png
CATEGORIES:ML MTP
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:DAYLIGHT
DTSTART:20250330T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR