BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Function estimation on large graphs with missing data - Alisa Kiri
 chenko (Universiteit van Amsterdam)
DTSTART:20160725T103000Z
DTEND:20160725T110000Z
UID:TALK66837@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Co-author: Harry van Zanten (University of Amsterdam)<span><br
 ><br>There are various problems in statistics and machine learning that in
 volve making an inference about a function on a graph. I will present a Ba
 yesian approach to estimating a smooth function in the context of regressi
 on and classification problems on graphs. I will discuss the mathematical 
 framework that allows to study the performance of nonparametric function e
 stimation methods on large graphs. I will review theoretical results that 
 show how to achieve asymptotically optimal Bayesian regularization under g
 eometry conditions on the families of the graphs and the smoothness assump
 tion on the true function. Both assumptions are formulated in terms of gra
 ph Laplacian. I will also discuss the case of "uniformly distributed" miss
 ing observations and investigate the generalization performance for variou
 s missing mechanisms.</span>
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR