BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Random Feature Expansions for Deep Gaussian Processes - Maurizio F
 ilippone (EURECOM)
DTSTART:20180206T100000Z
DTEND:20180206T110000Z
UID:TALK99934@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Drawing meaningful conclusions on the way complex real life ph
 enomena work and being able to predict the behavior of systems of interest
  require developing accurate and highly interpretable mathematical models 
 whose parameters need to be estimated from observations. In modern applica
 tions\, however\, we are often challenged with the lack of such models\, a
 nd even when these are available they are too computational demanding to b
 e suitable for standard parameter optimization/inference methods. While pr
 obabilistic models based on Deep Gaussian Processes (DGPs) offer attractiv
 e tools to tackle these challenges in a principled way and to allow for a 
 sound quantification of uncertainty\, carrying out inference for these mod
 els poses huge computational challenges that arguably hinder their wide ad
 option. In this talk\, I will present our contribution to the development 
 of practical and scalable inference for DGPs\, which can exploit distribut
 ed and GPU computing. In particular\, I will introduce a formulation of DG
 Ps based on random features that we infer using stochastic variational inf
 erence. Through a series of experiments\, I will illustrate how our propos
 al enables scalable deep probabilistic nonparametric modeling and signific
 antly advances the state-of-the-art on inference methods for DGPs.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR