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:Peter Whittle Lecture
SUMMARY:When Statistics Meets Computing - Professor Tony C
ai (Wharton\, University of Pennsylvania)
DTSTART;TZID=Europe/London:20181010T170000
DTEND;TZID=Europe/London:20181010T180000
UID:TALK108859AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/108859
DESCRIPTION:In the conventional statistical framework\, the go
al is developing optimal inference procedures\, wh
ere optimality is understood with respect to the s
ample size and parameter space. When the dimension
ality of the data becomes large as in many contemp
orary applications\, the computational concerns as
sociated with the statistical procedures come to t
he forefront. A fundamental question is: Is there
a price to pay for statistical performance if one
only considers computable (polynomial-time) proced
ures? After all\, statistical methods are useful i
n practice only if they can be computed within a r
easonable amount of time. \n\nIn this talk\, we di
scuss the interplay between statistical accuracy a
nd computational efficiency in two specific proble
ms: submatrix localization and sparse matrix detec
tion based on a noisy observation of a large matri
x. The results show some interesting phenomena tha
t are quite different from other high-dimensional
problems studied in the literature.\n\nA wine rece
ption in the central core will follow the talk.
LOCATION:Centre for Mathematical Sciences MR2
CONTACT:HoD Secretary\, DPMMS
END:VEVENT
END:VCALENDAR