AN APPROACH OF GROUPING DECOMPOSED COMPONENTS
Daisuke Ito, Takuya Mukai, Noboru Murata · 2003
In this paper, a grouping method by measuring the depen-dency of decomposed signals is discussed. Dependencies between components decomposed by some ICA algorithm are calculated using test statistics, based on the empirical characteristic function. This method helps selecting the com-bination of independent components by classifying them into subgroups. The relevance of these ideas is illustrated with synthesized data and auditory evoked MEG, and com-pared with the separability matrix method. 1.