BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:HAI Seminar with Erik Altman
X-WR-TIMEZONE:Pacific Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260608T080547Z
UID:tag:localist.com\,2008:EventInstance_49110386801961
DTSTART:20250507T190000Z
DTEND:20250507T201500Z
DESCRIPTION:IBM Synthetic Data Sets (SDS) have been created for use cases i
 n the financial industry.  One key focus is fraud and criminal activity\, 
 whose cost runs into the hundreds of billions of dollars per year or more.
   SDS labels many of these criminal activities including money laundering\
 , credit card fraud\, check fraud\, APP (Authorized Push Payment) fraud (s
 cams)\, and insurance claims fraud.  As such SDS data provides an attracti
 ve foundation for training AI detection models.\n\nUnlike much current act
 ivity around synthetic data generation\, SDS is not built using large lang
 uage models.  Instead SDS uses an agent-based virtual world approach.  A k
 ey advantage of the SDS design is that all labels are correct:  all fraud 
 is labelled fraud\, and only fraud is labelled fraud.  By contrast\, much 
 criminal activity is missed in the real world\, including 95% of money lau
 ndering by a UN estimate.  Hence\, even if real data is available\, it is 
 often of poor quality for training detection models\, or for generating sy
 nthetic data.\n\nIn practice\, access to real data is generally limited to
  a small number of people at the institution (e.g. a bank) that owns the d
 ata.  As such real data provides only a narrow view of activity at a singl
 e institution – as opposed to the global view provided by SDS data.  The
  SDS approach also yields a broad set of synthetic personal information.  
 This information is highly realistic despite using no information from rea
 l individuals.\n\nDevelopment of effective techniques for SDS has required
  deep expertise across diverse areas.  It has also required significant ma
 nual effort.  How to automate some of these efforts remains an open challe
 nge\, as do calibration\, scaling\, and other areas.
GEO:37.429987;-122.17333
LOCATION:Gates Computer Science Building\, 119
SUMMARY:HAI Seminar with Erik Altman
URL;VALUE=URI:https://events.stanford.edu/event/hai-seminar-with-erik-altma
 n
CATEGORIES:Class/Seminar
END:VEVENT
END:VCALENDAR