BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Approximate kernel embeddings of distributions - Dino Sejdinovic (
 University of Oxford)
DTSTART:20180501T100000Z
DTEND:20180501T110000Z
UID:TALK105094@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Kernel embeddings of distributions and the Maximum Mean Discre
 pancy (MMD)\, the resulting probability metric\, are useful tools for full
 y nonparametric hypothesis testing and for learning on distributional inpu
 ts\; i.e.\, where labels are only observed at an aggregate level. I will g
 ive an overview of this framework and describe the use of large-scale appr
 oximations to kernel embeddings in the context of Bayesian approaches to l
 earning on distributions and in the context of distributional covariate sh
 ift\; e.g.\, where measurement noise on the training inputs differs from t
 hat on the testing inputs.  <br><br><br><br>
LOCATION:Seminar Room 2\, Newton Institute
END:VEVENT
END:VCALENDAR