BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:iCalendar-Ruby
BEGIN:VEVENT
CATEGORIES:Academics,Campus & Community,Science & Tech
DESCRIPTION:Antonio Linero (Department of Statistics at Florida State Unive
rsity)\n\nTitle: Theory and Practice for Bayesian Regression Tree Ensembles
\n\nAbstract: Ensembles of decision trees have become a standard component
of the data analyst's toolkit\; commonly used algorithms include random for
ests and boosted decision trees. In this talk\, we investigate the properti
es of regression tree ensembles from a Bayesian standpoint. We focus on the
interplay between theory and practice to study the properties of ensembles
and obtain insights into (a) why decision tree ensembles are successful in
practice and (b) where they might be improved. We provide validation for t
he long-held hypothesis that BART ensembles perform well due to their abili
ty to detect low-order interactions\, a property which describes many real-
world settings. Further\, we identify two areas in which BART ensembles can
be expected to be suboptimal: under sparsity\, and when the underlying reg
ression function exhibits higher-order smoothness. We give theoretical supp
ort for these insights by establishing posterior contraction at near-optima
l rates adaptively across a large family of function spaces\, and provide e
mpirical support by applying our methodology to benchmark datasets. We conc
lude by presenting extensions of our methodology which account for other in
teresting structures beyond sparsity and smoothness\, and discuss how the i
nsights we obtain can be extended to non-Bayesian decision tree ensembling
methods.
DTEND:20190204T210000Z
DTSTAMP:20240814T224025Z
DTSTART:20190204T200000Z
GEO:30.28886;-97.739136
LOCATION:Burdine Hall (BUR)\, 136
SEQUENCE:0
SUMMARY:SDS Seminar Series - Antonio Linero
UID:tag:localist.com\,2008:EventInstance_4272436
URL:https://calendar.utexas.edu/event/sds_seminar_series_-_antonio_linero_7
213
END:VEVENT
END:VCALENDAR