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:Machine Learning @ CUED
SUMMARY:Gaussian processes\, spectral analysis kernels and
optimal transport - Felipe Tobar\, Universidad de
Chile
DTSTART;TZID=Europe/London:20221122T110000
DTEND;TZID=Europe/London:20221122T120000
UID:TALK193046AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/193046
DESCRIPTION:Gaussian processes (GPs) are Bayesian nonparametri
c generative models for time series and are partic
ularly well suited for continuous-time nonlinear r
egression tasks. The talk will start with a brief
introduction to GPs so as to illustrate their adva
ntages and challenges as well as to motivate their
use in a variety of tasks involving missing or ir
regularly-sampled data. We will then interpret the
GP model from a (Fourier) spectral analysis persp
ective and motivate the construction of covariance
functions based on the GPs frequency representati
on\; we will also show how GPs can be used for Spe
ctral Estimation. Then\, we will present recent ad
vances using optimal transport (a distance between
probability distributions) to define a distance b
etween GPs and explore alternative\, cost-efficien
t\, training strategies for GP. Throughout the tal
k\, we will show illustrative and real world examp
les. \n\nBio: Felipe is an Associate Professor at
the Initiative for Data and Artificial Intelligenc
e\, Universidad de Chile\, and the Director of the
Initiative for Data and Artificial Intelligence a
t the same institution. He holds Researcher positi
ons at the Center for Mathematical Modeling and th
e Advanced Center for Electrical and Electronic En
gineering. Prior to joining Universidad de Chile\,
Felipe was a postdoc at the Machine Learning Grou
p\, University of Cambridge\, during 2015 and he r
eceived a PhD in Signal Processing from Imperial C
ollege London in 2014. Felipe's research interests
lie in the interface between Machine Learning and
Statistical Signal Processing\, including approxi
mate inference\, Bayesian nonparametrics\, spectra
l estimation\, optimal transport and Gaussian proc
esses.
LOCATION:CBL Seminar Room
CONTACT:Dr R.E. Turner
END:VEVENT
END:VCALENDAR