BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Statistics
SUMMARY:Statistical inference for compound regression - Al
exandre Tsybakov\, Université Paris VI
DTSTART;TZID=Europe/London:20120518T160000
DTEND;TZID=Europe/London:20120518T170000
UID:TALK38172AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/38172
DESCRIPTION:This talk considers a general nonparametric regres
sion model called the compound model. It contains
as two special cases sparse additive regression an
d nonparametric regression with many covariates bu
t possibly a small number of relevant covariates.
A compound model is characterized by three main pa
rameters: the "microscopic" sparsity parameter ind
icating the number of relevant covariates\, the st
ructure parameter\, i.e.\, a binary sequence descr
ibing the "macroscopic" structure of the compound
function and the usual smoothness parameter corres
ponding to the complexity of the members of the co
mpound. We find non-asymptotic minimax rate of co
nvergence of estimators in such a model as a funct
ion of these three parameters. We also show that t
his rate can be attained in an adaptive way. This
is a joint work with Arnak Dalalyan and Yuri Ingst
er.
LOCATION:MR5\, CMS\, Wilberforce Road\, Cambridge\, CB3 0WB
CONTACT:Richard Samworth
END:VEVENT
END:VCALENDAR