BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Drupal iCal API//EN
X-WR-CALNAME:Events items teaser
X-WR-TIMEZONE:America/Toronto
BEGIN:VTIMEZONE
TZID:America/Toronto
X-LIC-LOCATION:America/Toronto
BEGIN:DAYLIGHT
TZNAME:EDT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
DTSTART:20240310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZNAME:EST
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
DTSTART:20241103T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:68276e949f5a8
DTSTART;TZID=America/Toronto:20250214T153000
SEQUENCE:0
TRANSP:TRANSPARENT
DTEND;TZID=America/Toronto:20250214T163000
URL:/combinatorics-and-optimization/events/tutte-colloq
 uium-xi-he
SUMMARY:Tutte colloquium-Xi He
CLASS:PUBLIC
DESCRIPTION:Summary \n\nTITLE:Accuracy Aware Minimally Invasive Data Explor
 ation For Decision\nSupport\n\nSPEAKER:\n Xi He\n\nAFFILIATION:\n Universi
 ty of À¶Ý®ÊÓƵ\n\nLOCATION:\n MC 5501\n\nABSTRACT: Decision-support (DS) 
 applications\, crucial for timely and\ninformed decision-making\, often an
 alyze sensitive data\, raising\nsignificant privacy concerns. While privac
 y-preserving randomized\nmechanisms can mitigate these concerns\, they int
 roduce the risk of\nboth false positives and false negatives. Critically\,
  in DS\napplications\, the number of false negatives often needs to be str
 ictly\ncontrolled. Existing privacy-preserving techniques like differentia
 l\nprivacy\, even when adapted\, struggle to meet this requirement without
 \nsubstantial privacy leakage\, particularly when data distributions are\n
 skewed. This talk introduces a novel approach to minimally invasive\ndata 
 exploration for decision support. Our method minimizes privacy\nloss while
  guaranteeing a bound on false negatives by dynamically\nadapting privacy 
 levels based on the underlying data distribution. We\nfurther extend this 
 approach to handle complex DS queries\, which may\ninvolve multiple condit
 ions on diverse aggregate statistics combined\nthrough logical disjunction
  and conjunction. Specifically\, we define\ncomplex DS queries and their a
 ssociated accuracy requirements\, and\npresent algorithms that strategical
 ly allocate a privacy budget to\nminimize overall privacy loss while satis
 fying the bounded accuracy\nguarantee.\n\n \n\n \n 
DTSTAMP:20250516T165756Z
END:VEVENT
END:VCALENDAR