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:An Automated Statistician which learns Bayesian no
nparametric models of time series data - Ghahraman
i\, Z (University of Cambridge)
DTSTART;TZID=Europe/London:20140116T141500
DTEND;TZID=Europe/London:20140116T150000
UID:TALK49977AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/49977
DESCRIPTION:I will describe the "Automated Statistician"\, a p
roject which aims to automate the exploratory anal
ysis and modelling of data. Our approach starts by
defining a large space of related probabilistic m
odels via a grammar over models\, and then uses Ba
yesian marginal likelihood computations to search
over this space for one or a few good models of th
e data. The aim is to find models which have both
good predictive performance\, and are somewhat int
erpretable. Our initial work has focused on the le
arning of unknown nonparametric regression functio
ns\, and on learning models of time series data\,
both using Gaussian processes. Once a good model h
as been found\, the Automated Statistician generat
es a natural language summary of the analysis\, pr
oducing a 10-15 page report with plots and tables
describing the analysis. I will focus in particula
r on the modelling of time series\, including how
we handle change points in Gaussian process models
. I will also discuss challenges su ch as: how to
trade off predictive performance and interpretabil
ity\, how to translate complex statistical concept
s into natural language text that is understandabl
e by a numerate non-statistician\, and how to inte
grate model checking.\n \nThis is joint work with
James Lloyd and David Duvenaud (Cambridge) and Rog
er Grosse and Josh Tenenbaum (MIT).\n
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:Mustapha Amrani
END:VEVENT
END:VCALENDAR