BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Variational Inference in Gaussian Processes for non-linear time se
 ries - Carl Edward Rasmussen\, University of Cambridge
DTSTART:20160204T140000Z
DTEND:20160204T150000Z
UID:TALK63210@talks.cam.ac.uk
CONTACT:Tim Hughes
DESCRIPTION:Modelling non-linear dynamical systems from observations (syst
 em identification) is often a pre-requisite for designing good controllers
 . Despite the fundamental importance of this problem\, good models and inf
 erence methods remain a technical challenge. The difficulties include havi
 ng to deal with noisy and incomplete measurements and treating suitably fl
 exible non-linear models. Sufficiently flexible models generally won’t h
 ave succinct representations\, and large numbers of free parameters requir
 es principled approaches to issues such as overfitting. In this talk I wil
 l present a couple of recent developments in machine learning which togeth
 er allows approximate fully Bayesian inference jointly over both latent st
 ates and non-linear transition models. This elegant and practical framewor
 k relies on variational inference in non-linear Gaussian process state spa
 ce models.
LOCATION:Cambridge University Engineering Department\, LR5
END:VEVENT
END:VCALENDAR