BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:Europe/Vienna
BEGIN:DAYLIGHT
DTSTART:20260329T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251026T020000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260415T110113Z
UID:698f1a0aaf212263034654@ist.ac.at
DTSTART:20260226T113000
DTEND:20260226T123000
DESCRIPTION:Speaker: Nikita Kalinin\nAbstract: The lower-triangular all-one
 s counting matrix is a key primitive in differential privacy\, and its fac
 torization norms determine theoretical utility guarantees for private mach
 ine learning training with correlated noise. For more than three decades\,
  the best known upper bound for these norms remained essentially unchanged
 \, and recent work asked whether an explicit factorization could provably 
 improve it. In this talk I present an explicit\, efficiently computable co
 nstruction that improves the longstanding bound\, together with significan
 tly stronger lower bounds\, shrinking the remaining constant gap to a smal
 l margin.
LOCATION:Office Bldg West / Ground floor / Foyer seminar room (I21.EG.128)\
 , ISTA
ORGANIZER:achaturv@ist.ac.at
SUMMARY:Nikita Kalinin: TCS Seminar - Sharper Matrix Factorization Bounds f
 or Differentially Private Continual Counting
URL:https://talks-calendar.ista.ac.at/events/6298
END:VEVENT
END:VCALENDAR