Single Trial ERP detection Exploiting Sparsity in Time

Foad Ghaderi

In: Proceedings of the International Biomedical and Astronomical Signal Processing (BASP) Frontiers Workshop. The international biomedical and astronomical signal processing (BASP) Frontiers workshop (BASP-2013) January 27-February 1 Villars-sur-Ollon Switzerland Seite 62 2/2013.


Sparsity of event related potentials (ERP) in time domain is used to develop a single trial ERP detection method. It is assumed that ERPs are deterministic whereas the ongoing EEG does not have this property [1]. Using the recursive projected compressive sensing (Re-ProCS) method [2] the electroencephalogram (EEG) data is decomposed into a high dimensional, low rank component and a sparse component which represents the ERPs. The method shows promising results on EEG data which is synthetically generated using real EEG and ERP samples.


Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence