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:Inference in generative models using the Wasserste
in distance - Christian Robert (CNRS &\; Univer
sitÃ© Paris-Dauphine )
DTSTART;TZID=Europe/London:20170707T114500
DTEND;TZID=Europe/London:20170707T123000
UID:TALK73186AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/73186
DESCRIPTION:In purely generative models\, one can simulate dat
a given parameters but not necessarily evaluate th
e likelihood. We use Wasserstein distances between
empirical distributions of observed data and empi
rical distributions of synthetic data drawn from s
uch models to estimate their parameters. Previous
interest in the Wasserstein distance for statistic
al inference has been mainly theoretical\, due to
computational limitations. Thanks to recent advanc
es in numerical transport\, the computation of the
se distances has become feasible\, up to controlla
ble approximation errors. We leverage these advanc
es to propose point estimators and quasi-Bayesian
distributions for parameter inference\, first for
independent data. For dependent data\, we extend t
he approach by using delay reconstruction and resi
dual reconstruction techniques. For large data set
s\, we propose an alternative distance using the H
ilbert space-filling curve\, which computation sca
les as* **n log n** *where n is the
size of the data. We provide a theoretical study
of the proposed estimators\, and adaptive Monte Ca
rlo algorithms to approximate them. The approach i
s illustrated on four examples: a quantile g-and-k
distribution\, a toggle switch model from systems
biology\, a Lotka-Volterra model for plankton pop
ulation sizes and a L\\'\;evy-driven stochastic
volatility model.

*[This is joint wor
k with **Espen Bernton (Harvard Uni
versity)\, Pierre E. Jacob (Harvard University)\
, Mathieu Gerber (University of Bristol).]*<
/b>

LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR