BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Secure active causal dataset acquisition - Desi Ivanova (Universit
 y of Oxford)
DTSTART:20250624T081500Z
DTEND:20250624T091500Z
UID:TALK232198@talks.cam.ac.uk
DESCRIPTION:Merging datasets across institutions is a lengthy and costly p
 rocedure\, especially when it involves private information. Data hosts may
  therefore want to prospectively gauge which datasets are most beneficial 
 to merge with\, without revealing sensitive information. For causal estima
 tion this is particularly challenging as the value of a merge will depend 
 not only on the reduction in epistemic uncertainty but also the improvemen
 t in overlap. To address this challenge\, we introduce the first cryptogra
 phically secure information-theoretic approach for quantifying the value o
 f a merge in the context of heterogeneous treatment effect estimation. We 
 do this by evaluating the Expected Information Gain (EIG) and utilising mu
 lti-party computation to ensure it can be securely computed without reveal
 ing any raw data. As we demonstrate\, this can be used with differential p
 rivacy (DP) to ensure privacy requirements whilst preserving more accurate
  computation than naive DP alone. To the best of our knowledge\, this work
  presents the first privacy-preserving method for dataset acquisition tail
 ored to causal estimation. We demonstrate the effectiveness and reliabilit
 y of our method on a range of simulated and realistic benchmarks.&nbsp\;
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR