Source Separation and Clustering of Phase-Locked Subspaces
Almeida, M.
; Schleimer, J.-H.
; Vigário, R. V.
;
Bioucas-Dias, J.
IEEE Transactions on Neural Networks Vol. 22, Nº 9, pp. 1419 - 1434, September, 2011.
ISSN (print): 1045-9227
ISSN (online): 1941-0093
Scimago Journal Ranking: (in )
Digital Object Identifier: 10.1109/TNN.2011.2161674
Download Full text PDF ( 1 MB)
Abstract
It has been proven that there are synchrony (or phase-locking) phenomena present in multiple oscillating systems such as electrical circuits, lasers, chemical reactions and human neurons. If the measurements of these systems cannot detect the individual oscillators but rather a superposition of them, as in brain electrophysiological signals (EEG and MEG), spurious phase-locking will be detected. Current source-extraction techniques attempt to undo this superposition by assuming properties on the data which are not valid when underlying sources are phase-locked. Statistical independence of the sources is one such invalid assumption, as phase-locked sources are dependent. In this article we introduce methods for source separation and clustering which make adequate assumptions for data where synchrony is present, and show with simulated data that they perform well even in cases where ICA and other well-known source-separation methods fail. The results in this paper provide a proof-of-concept that synchrony-based techniques are useful for low-noise applications.