BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Kernel Mean Embeddings - Elre Oldewage (University of Cambridge)
DTSTART:20200219T110000Z
DTEND:20200219T123000Z
UID:TALK139993@talks.cam.ac.uk
CONTACT:75379
DESCRIPTION:Kernel mean embeddings are a technique that represents probabi
 lity distributions as elements in a high dimensional Hilbert space. This a
 llows difficult operations such as expectations to be reformulated as dot 
 products in the Hilbert space.\nThis talk introduces reproducing kernel Hi
 lbert spaces (RKHS)\, which is the theoretical underpinnings upon which ke
 rnel mean embeddings are constructed. The talk aims to provide intuition r
 egarding the relationship between kernels\, feature maps and RHKS function
  spaces. The talk continues with a discussion of kernel mean embeddings\, 
 and common applications of kernel mean embeddings\, specifically\, kernel 
 two-sample hypothesis testing. We also discuss embedding multivariable con
 ditional distributions which allow the application of kernel Bayes rule.
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR