BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Debiased Whittle likelihood for time series and spatial data
DTSTART;VALUE=DATE-TIME:20211022T151500
DTEND;VALUE=DATE-TIME:20211022T170000
UID:9bcd0c354480d1e46af1ed695abef51b8edcfeff56bfa53802c11355
CATEGORIES:Conferences - Seminars
DESCRIPTION:Arthur Guillaumin\, School of Mathematical Sciences\, Universi
ty of London\nTime series and spatial data are ubiquitous in many applicat
ion areas\, such as environmental data\, geosciences\, astronomy\, and fin
ance. A key statistical modelling and estimation challenge for these data
is that of dependance between points at different times or locations. Whil
e parametric models of covariance can be estimated via exact likelihood\,
this is ill-suited for many practical problems due to the heavy computatio
nal cost.\n\nA standard approach to address this relies on approximate lik
elihood methods. The Whittle likelihood is one such approximation for grid
ded data\, based on the Discrete Fourier Transform of the data. It is popu
lar due to its n log n computational cost\, robustness to non-Gaussian dat
a\, and amenability to interpretation in the spectral domain. However\, Wh
ittle likelihood estimates suffer from a strong bias due to the finite and
discrete sampling. This is true in particular for spatial data where bias
dominates verses standard deviation in dimension equal or greater than tw
o. Additionally\, practical sampling patterns often diverge from theoretic
al requirements\, due to non-square observational domains or missing data.
In this presentation we present a recently proposed modification to the W
hittle likelihood which addresses all these issues at once.\nWe provide as
ymptotic results under a framework which we call Significant Correlation C
ontribution\, which allows us to understand the interplay between the samp
ling pattern and the covariance model. We demonstrate that our modificatio
n renders our estimate asymptotically efficient and normal for a wide clas
s of settings. \n
LOCATION:https://epfl.zoom.us/j/66227383868
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR