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:Isaac Newton Institute Seminar Series
SUMMARY:Strong converses and high-dimensional statistical
estimation problems - Ramji Venkataramanan (Univer
sity of Cambridge)
DTSTART;TZID=Europe/London:20180724T094500
DTEND;TZID=Europe/London:20180724T103000
UID:TALK108283AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/108283
DESCRIPTION:In many statistical inference problems\, we
wish to bound the performance of any possible esti
mator. This can be seen as a converse result\, in
a standard information-theoretic sense. A standard
approach in the statistical literature is based o
n Fano&rsquo\;s inequality\, which typically gives
a weak converse. We adapt these arguments by repl
acing Fano by more recent information-theoretic id
eas\, based on the work of Polyanskiy\, Poor and V
erdu. This gives tighter lower bounds that can be
easily computed and are asymptotically sharp. We i
llustrate our technique in three applications: den
sity estimation\, active learning of a binary clas
sifier\, and compressed sensing\, obtaining tighte
r risk lower bounds in each case.  \;
(joint with Oliver Johnson\, see doi:10.1214/18-
EJS14)
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR