BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Leveraging Black-box Models to Assess Feature Importance - Jing Zh
 ou (University of East Anglia)
DTSTART:20250508T130000Z
DTEND:20250508T133000Z
UID:TALK230485@talks.cam.ac.uk
DESCRIPTION:Understanding the impact of changes in features on the uncondi
 tional distribution of outcomes is crucial for various applications. Despi
 te their predictive accuracy\, existing black-box models are limited in ad
 dressing such questions. In this work\, we propose a novel approximation m
 ethod to compute feature importance curves\, which quantify changes across
  the quantiles of the outcome distribution due to shifts in features. Our 
 approach leverages pre-trained black-box models\, combining their predicti
 ve strength with interpretation. Through extensive simulations and real-wo
 rld data applications\, we show that our method delivers sparse\, reliable
  results while maintaining computational efficiency\, making it a practica
 l tool for interpretation.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR