San Francisco Bay Area Statistical Association
Meeting
Speaker |
Alexandra
Piryatinska, Assistant Professor in the Department
of Mathematics at San Francisco State University. |
Title |
Change-point
detection for non-stationary time series via complexity of functions and
its applications to EEG data |
Time |
Thursday,
Nov. 8 at 4:30pm. |
Location |
San
Francisco State University, 1600 Holloway Avenue, Thornton Hall 404 http://www.sfsu.edu/~sfsumap/ and general
parking/driving/public transit directions to campus are linked here - http://www.sfsu.edu/~parking/directions/ |
Abstract:
In many applications time
series are sequences of connected, distinct segments which are
generated by their own individual mechanisms. To analyze such series it
is necessary to split them into these segments. If time series is generated by
stochastic mechanisms, then the classical change-point detection algorithm can
achieve the segmentation. However it is not the case for deterministic or mixed
mechanisms.
We propose a novel approach to this
problem based on a new concept of the complexity of a continuous function. We
show that the dependence of the complexity of a function on the reconstruction
error can be well approximated in logarithmic coordinates by an affine
function. Its parameters, calculated dynamically, are then used as
diagnostic sequences to find the change-points of the original time
series. We verify the effectiveness of this procedure in the case of
several simulated time series and apply this approach to the EEG data. (Joint
work with B. Darkovsky)
Return
to Bay
Area ASA Homepage