Unbiased detection of phase locking in multivariate signals

Axel Hutt, Andreas Daffertshofer, and Ulrich Steinmetz
(Please use for correspondence this email).

Submission date: 23. Sep. 2002
paper submitted to: Physics Letters A
MSC-Numbers: 37H10, 58C99
PACS-Numbers: 02.50.Sk, 05.45.Xt, 05.10.-a
Keywords and phrases: clustering, segmentation, multivariate analysis

Abstract:
We present a cluster method for the detection of meta-stable states in multivariate periodic time series. In due to the topology of periodic data, we formulate a K-Means clustering. The method is illustrated using simulated multivariate phase dynamics, in which temporal segments of phase-locked states are registered.