Description of the method.
Authors: A. Schloegl, C. Vidaurre
Downsampling to 128Hz
- extract each possible bipolar channels (496) + 32 monopolar channel. From each channel extract the following parameters:
adaptive estimates of autoregressive model parameters with non-zero mean. A model order of p = 3 is used. Alpha (10-12Hz) and Beta (16-24Hz) bandpower The parameters AAR(3)+adaptive mean+BP(alpha)+BP(beta) of each single were used to estimated a statistical classifier. the channel with the best separability was selected
SUBJ 1: #28 - #10
SUBJ 2: #10 - #2
SUBJ 3: #26 (monopolar)
To calculate the classification on the test set, only the parameters of the best channels were classified using linear discriminant analysis.