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:Third and Fourth Cumulants in Independent Componen
t Analysis - Hannu Oja\, University of Turku
DTSTART;TZID=Europe/London:20150522T160000
DTEND;TZID=Europe/London:20150522T170000
UID:TALK58613AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/58613
DESCRIPTION:The independent component model is a latent varia
ble model where the components of the observed ran
dom vectors are linear combination of latent inde
pendent variables. The aim is then to find an esti
mate for a transformation matrix back to independe
nt components. In this talk\, we consider the join
t use of third and fourth cumulants in finding ind
ependent components. First\, univariate cumulants
are used as projection indices in search for indep
endent components (projection pursuit). Second\, m
ultivariate cumulant matrices are jointly used to
solve the problem. The properties of the estimate
s are considered in detail through corresponding o
ptimization problems\, estimating equations\, alg
orithms and asymptotic statistical properties. Co
mparisons of the asymptotic variances of the estim
ates in wide independent component models show tha
t in most cases projection pursuit approach using
both third and fourth scared cumulants is a safe c
hoice.
LOCATION:MR12\, Centre for Mathematical Sciences\, Wilberf
orce Road\, Cambridge
CONTACT:
END:VEVENT
END:VCALENDAR