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:A compositional approach to scalable statistical m
odelling and computation - Darren Wilkinson (Newca
stle University)
DTSTART;TZID=Europe/London:20180208T110000
DTEND;TZID=Europe/London:20180208T120000
UID:TALK100723AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/100723
DESCRIPTION:In statistics\, and in life\, we typically solve b
ig problems by (recursively) breaking them down in
to smaller problems that we can solve more easily\
, and then compose the solutions of the smaller pr
oblems to provide a solution to the big problem th
at we are really interested in. This "divide and c
onquer" approach is necessary for the development
of genuinely scalable models and algorithms. It is
therefore unfortunate that statistical models and
algorithms are not usually formulated in a compos
able way\, and that the programming languages typi
cally used for scientific and statistical computin
g also fail to naturally support composition of mo
dels\, data and computation. The mathematical subj
ect of category theory is in many ways the mathema
tical study of composition\, and provides signific
ant insight into the development of more compositi
onal models of computation. Functional programming
languages which are strongly influenced by catego
ry theory turn out to be much better suited to the
development of scalable statistical algorithms th
an the imperative programming languages more commo
nly used. Expressing algorithms in a functional/ca
tegorical way is not only more elegant\, concise a
nd less error-prone\, but provides numerous more t
angible benefits\, such as automatic parallelisati
on and distribution of algorithms. I will illustra
te the concepts using examples such as the statist
ical analysis of streaming data\, image analysis\,
numerical integration of PDEs\, particle filterin
g\, Gibbs sampling\, and probabilistic programming
\, using concepts from category theory such as fun
ctors\, monads and comonads. Illustrative code sni
ppets will given using the Scala programming langu
age.
LOCATION:Seminar Room 2\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR