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:(CANCELLED) Stochastic Causal Programming for Boun
ding Treatment Effects - Ricardo Silva (UCL)
DTSTART;TZID=Europe/London:20220617T140000
DTEND;TZID=Europe/London:20220617T150000
UID:TALK173327AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/173327
DESCRIPTION:Causal effect estimation is important for numerous
tasks in the natural and social sciences. However
\, identifying effects is impossible from observat
ional data without making strong\, often untestabl
e assumptions. We consider algorithms for the part
ial identification problem\, bounding treatment ef
fects from multivariate\, continuous treatments ov
er multiple possible causal models when unmeasured
confounding makes identification impossible. We c
onsider a framework where observable evidence is m
atched to the implications of constraints encoded
in a causal model by norm-based criteria. This gen
eralizes classical approaches based purely on gene
rative models. Casting causal effects as objective
functions in a constrained optimization problem\,
we combine flexible learning algorithms with Mont
e Carlo methods to implement a family of solutions
under the name of stochastic causal programming.
In particular\, we present ways by which such cons
trained optimization problems can be parameterized
without likelihood functions for the causal or th
e observed data model\, reducing the computational
and statistical complexity of the task.\n \nJoint
work with Kirtan Padh\, Jakob Zeitler\, David Wat
son\, Matt Kusner and Niki Kilbertus
LOCATION:MR12\, Centre for Mathematical Sciences
CONTACT:Qingyuan Zhao
END:VEVENT
END:VCALENDAR