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:Recent Developments in the Study of Single-Index T
ype Models - Mouli Banerjee\, University of Michig
an
DTSTART;TZID=Europe/London:20190607T140000
DTEND;TZID=Europe/London:20190607T150000
UID:TALK119416AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/119416
DESCRIPTION:Single-index type models are popular in statistics
\, biostatistics and economics as they alleviate t
he curse of dimensionality to a considerable exten
t while allowing for broad classes of models throu
gh the dependence on an unknown link function. Var
ious classes of single index models are known: in
regular models under a fixed dimension setting [i.
e. fixed number of regression parameters]\, the re
gression parameter is √n estimable\; under current
status type censoring of the response variable in
a linear regression model — which leads to the bi
nary choice model — one gets a generalized single-
index model where the rate of estimation is at mos
t n^{1/3}\, a problem well-studied in the economet
rics literature (by Manski and subsequent authors)
. Single index structures with a discontinuous lin
k function arise in models involving change-planes
in multidimensional space — hyperplanes that sepa
rate two (or more) response or survival regimes —
and are relevant to applications in personalized m
edicine and dynamic treatment regimes. Here\, rate
s of estimation can easily exceed √n in the finite
dimensional case. \n\nI will talk about some of
my recent work in the above class of models in gro
wing and high dimensional settings focusing on how
growing dimensions introduce significantly new ch
allenges at both theoretical and computational lev
els and present some recent results on convergence
\, minimax-optimal rates\, and inference\, as well
as future challenges. \n\nThe talk is based on j
oint work with Ya'acov Ritov\, Hamid Eftekhari\, Z
hiyuan Lu and Debarghya Mukherjee.
LOCATION:MR12
CONTACT:Dr Sergio Bacallado
END:VEVENT
END:VCALENDAR