BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Consistent Validation for Predictive Methods in Spatial Settings -
  David Burt\, MIT
DTSTART:20240223T130000Z
DTEND:20240223T140000Z
UID:TALK212503@talks.cam.ac.uk
CONTACT:Dr R.E. Turner
DESCRIPTION:Spatial prediction tasks are key to weather forecasting\, stud
 ying air pollution\, and other scientific endeavors. Determining how much 
 to trust predictions made by statistical or physical methods is essential 
 for the credibility of scientific conclusions. Unfortunately\, classical a
 pproaches for validation fail to handle mismatch between locations availab
 le for validation and (test) locations where we want to make predictions. 
 This mismatch is often not an instance of covariate shift (as commonly for
 malized) because the validation and test locations are fixed (e.g.\, on a 
 grid or at select points) rather than i.i.d. from two distributions. In th
 e present work\, we formalize a check on validation methods: that they bec
 ome arbitrarily accurate as validation data becomes arbitrarily dense. We 
 show that classical and covariate-shift methods can fail this check. We in
 stead propose a method that builds from existing ideas in the covariate-sh
 ift literature\, but adapts them to the validation data at hand. We prove 
 that our proposal passes our check. And we demonstrate its advantages empi
 rically on simulated and real data.
LOCATION:Cambridge University Engineering Department\, CBL Seminar room BE
 4-38.
END:VEVENT
END:VCALENDAR