Title :
Joint spatial and spectral filter estimation for single trial detection of event related potentials
Author_Institution :
Robot. Group, Univ. of Bremen, Bremen, Germany
Abstract :
It is usually assumed that the main frequency components of ERPs are in some specific frequency bands. Therefore, predefined cutoff frequencies are used to filter the raw data in different applications. Here, an extension of the periodic spatial filter method is proposed that jointly learns spatial filters and the corresponding FIR filters to be used for single trial ERP detection. Experimental results confirm that the proposed spatio-spectral filtering method outperforms its predecessor spatial filter.
Keywords :
FIR filters; bioelectric potentials; electroencephalography; medical signal processing; optical filters; spatial filters; ERP detection; FIR filters; electroencephalogram; event related potentials; frequency bands; frequency components; periodic spatial filter method; predefined cutoff frequencies; raw data filter; single trial detection; spatial filters; spatio-spectral filtering; spectral filter estimation; Covariance matrices; Cutoff frequency; Eigenvalues and eigenfunctions; Electroencephalography; Finite impulse response filters; Noise; Vectors; ERP detection; Electroencephalogram; periodic spatial filter; spatio-spectral filters;
Conference_Titel :
Machine Learning for Signal Processing (MLSP), 2013 IEEE International Workshop on
Conference_Location :
Southampton
DOI :
10.1109/MLSP.2013.6661938