MEG Reading Group
Space-by-time decomposition of EEG data
after the usual Monday MEG meeting (6th Floor Meeting Room; 11am)
"Space-by-time decomposition of EEG data".
Abstract: I will present a model (named space-by-time decomposition) that factorizes neuroimaging data into spatial and temporal components. The model is implemented using an algorithm (sNM3F) based on Non-Negative Matrix Factorization. To illustrate the approach, we apply it on EEG data collected during performance of a 2AFC categorization task (Philiastides & Sajda, 2006). We show that the space-by-time model provides a low-dimensional and task-relevant representation of EEG spatial and temporal structure. The extracted components have a well-characterized functional role and allow reliable single-trial stimulus categorization.