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:Random Planted Forest: a directly interpretable tr
ee ensemble - Enno Mammen (Heidelberg University)
DTSTART;TZID=Europe/London:20220304T140000
DTEND;TZID=Europe/London:20220304T150000
UID:TALK168119AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/168119
DESCRIPTION:We introduce a novel interpretable and tree-based
algorithm for prediction in a regression setting i
n which each tree in a classical random forest is
replaced by a family of planted trees that grow si
multaneously.\nThe motivation for our algorithm is
to estimate the unknown regression function from
a functional ANOVA decomposition perspective\, whe
re each tree corresponds to a function within that
decomposition.\nTherefore\, planted trees are lim
ited in the number of interaction terms. The maxi
mal order of approximation in the ANOVA decomposit
ion can be specified or left unlimited. If a first
order approximation is chosen\, the result is an
additive model. In the other extreme case\, if the
order of approximation is not limited\, the resul
ting model places no restrictions on the form of t
he regression function. In a simulation study we f
ind encouraging prediction and visualisation prope
rties of our random planted forest method.\nWe al
so develop theory for an idealised version of rand
om planted forests. In particular\, for an additiv
e model we show that the idealised version achieve
s asymptotically optimal one-dimensional convergen
ce rates of order $n^{-2/5}$ up to a logarithmic f
actor. The talk reports on joint work with Munir H
iabu (Copenhagen) and Joseph Theo Meyer (Heidelber
g).
LOCATION:https://maths-cam-ac-uk.zoom.us/j/93998865836?pwd=
VzVzN1VFQ0xjS3VDdlY0enBVckY5dz09
CONTACT:Qingyuan Zhao
END:VEVENT
END:VCALENDAR