BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Multi-output conformal regression - Souhaib Ben Taieb (Mohamed bin
  Zayed University of Artificial Intelligence)
DTSTART:20250604T104500Z
DTEND:20250604T110500Z
UID:TALK230806@talks.cam.ac.uk
DESCRIPTION:Conformal prediction allows us to construct distribution-free 
 prediction regions with finite-sample coverage guarantees. While well-esta
 blished in univariate settings\, its extension to multi-output regression 
 remains relatively underexplored and introduces additional challenges\, su
 ch as capturing complex output dependencies and increased computational co
 mplexity.\nIn this presentation\, we introduce a unified comparative study
  of several conformal methods for multi-output prediction\, including diff
 erent classes of conformity scores. By integrating these methods within a 
 common framework\, we highlight key methodological differences and identif
 y connections between them. We also compare their empirical performance ac
 ross a range of real-world tabular regression datasets.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR