BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Choosing a good histogram - Yannick Baraud (Univ.  Nice)
DTSTART:20091016T150000Z
DTEND:20091016T160000Z
UID:TALK20008@talks.cam.ac.uk
CONTACT:Richard Nickl
DESCRIPTION:Histograms are probably among the most\nsimple and popular est
 imators. They are\nwidely used in science\, especially\nby non-statisticia
 ns\, in view of estimating\ndensities (or intensities of point processes).
 \nTo build a good histogram\, one needs\nto partition the data in a suitab
 le way\nwhich turns out to be a tricky problem. Given a\n(possibly large) 
 family of candidate\npartitions\, how can we select a suitable one\non whi
 ch our histogram will be as close as\npossible to  the unknown density?\nB
 esides\, are there families of partitions\none should consider preferably?
  These\nare some of the questions we shall try\nto answer in this talk by 
 adopting a\nnon-asymptotic point of view.\n\n
LOCATION:MR12\, CMS\, Wilberforce Road\, Cambridge\, CB3 0WB
END:VEVENT
END:VCALENDAR