This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Slide 1
Amir Omidvarnia 3 June 2011 Synchrony measures for newborn EEG
analysis
Slide 2
Outline The concept of cointegration Johansen test Bivariate
phase synchronization Multivariate phase synchronization Surrogate
data method for phase synchronization Results White noise
Combination of white noise and random walk Asymm./Asynch. data
Slide 3
The concept of cointegration Suppose two drunkards are
wandering aimlessly, while they dont know each other. Their
movement can be considered as two independent random walks.
Mathematically, random walk refers to a trajectory that consists of
taking successive random steps.
Slide 4
The concept of cointegration Random walk
Slide 5
The concept of cointegration Bivariate cointegration: A drunk
and her dog [2] Now, imagine a drunk walking with her dog. Each of
the two trajectories is still a random walk by itself. But, the
distance between two paths is fairly predictable, as the location
of the one can roughly tell us the location of the other one.
Slide 6
The concept of cointegration Bivariate cointegration: A drunk
and her dog
Slide 7
The concept of cointegration Bivariate cointegration A long-run
equilibrium relationship between the drunk and her dog causes a
stationary distance between their random walks. The co-movement
between two random walks is called a bivariate cointegrating
relationship.
Slide 8
The concept of cointegration Bivariate cointegration x 1 (t), x
2 (t): Random walk -d < x 1 (t) + x 2 (t) < d x x 1 (t) ( /
)x 2 (t) ~ white noise [ ] : cointegrating vector
Slide 9
The concept of cointegration multivariate cointegration: A
sheep herd and Shepherd dogs [6] Imagine a herd of sheep wandering
aimlessly in the field (multiple random walks). Consider that a
herding dog guards the flock by running around and returning the
sheep that have gone too far back to the herd.
Slide 10
The concept of cointegration multivariate cointegration: A
sheep herd and Shepherd dogs The dog makes a co-movement among the
sheep. Thus, we can say that there is a cointegrating relationship
(rank of one) within the sheep trajectories. It is obvious that two
dogs (rank of two) are able to restrict the movements within the
flock more than one dog. The higher cointegration rank,The higher
cointegration rank, the more restricted movementsthe more
restricted movements
Slide 11
Johansen test A test for estimation of the cointegrating
vectors and co- movements within a K-channel signal. Two types of
hypothesis testing are applied on a mathematical feature extracted
from the signal: Trace test H 0 : There are r cointegrating
relations H 1 : There are K cointegrating relations Maximum
eigenvalue test H 0 : There are r cointegrating relations H 1 :
There are r+1 cointegrating relations.
Slide 12
Johansen test Johansen test gives all linear combinations
between the dimensions which represent a stationary random process.
Let x(t) be K-channel signal (t = 1,...,T). A cointegrating
relationship among the channels (out of r, r