Materials: [ Cód.: GPShannonEN.mlx ] [ PDF ]
This video presents the ‘statistical’ (Gaussian process) version of Shannon’s
sampling theorem. The discussion will be split into two videos. The first one will
only focus on motivation, reviewing of sampling theorem statements, presenting
the power spectral density and auto-covariance (sinc function) of a band-limited
stochastic process, and simulating realizations of such a process. The video
concludes with a motivational presentation of the principal component analysis on
a finite grid of points in the time domain: a 30-second fragment realization
(31 samples, every second) gridded at 0.05 seconds can be described as
a realization of just 34 independent random variables... We are almost
there: we have 31 statistically independent samples... when the number
of samples grows to infinity, we reach perfect reconstruction from such
samples. Details and Matlab examples will be fully covered in the next video:
[
*Link to my [ whole collection] of videos in English. Link to larger [ Colección completa] in Spanish.
Colección completa en ESPA
Previous Matérn kernel regression (2): 2nd 3rd and fractional order, splines, Gaussian filter examples
Next Shannon sampling theorem: Gaussian process version (2)
Full collection in ENGLISH [View all]:
Previous Matérn kernel regression (2): 2nd 3rd and fractional order, splines, Gaussian filter examples
Next Shannon sampling theorem: Gaussian process version (2)