Normalizer and Expression Filter
Posted: 29 Aug 2014, 06:22
Dear all,
I am implementing a standard Motor Imagery paradigm for cursor control in 2D, where the cursor moves at constant speed on x axis and amplitude in y direction is controlled by the spectral power changes at C3 (one only feature, one only output channel from LinearClassifier).
Now, I have two questions:
1) I would like to set Expression Filter to give me the dB of the spectral power, but if I set it to 1x1 matrix with value "-10*log10(Signal(1,1))" the Operator returns me the error: "ExpressionFilter::Preflight: State "log10" is inaccessible".
2) I really can not get how to configure Normalizer! In the case I would like NOT to have adaptation I should set Adaptation to 0 for each of the ExpressionFilter's output channel (only one, in this case, right?) and offsets and gains could have any value because they will not be taken into account. Am I right? But, if I would like to HAVE adaptation (in mean and variance both) WHICH and HOW MANY parameters should I set at the beginning of the experiment? Is the number of parameters equal to the number of control signals, e.g. only one in this case (the one that controls the y-direction movement)??
Thank you in advance,
Giulia
I am implementing a standard Motor Imagery paradigm for cursor control in 2D, where the cursor moves at constant speed on x axis and amplitude in y direction is controlled by the spectral power changes at C3 (one only feature, one only output channel from LinearClassifier).
Now, I have two questions:
1) I would like to set Expression Filter to give me the dB of the spectral power, but if I set it to 1x1 matrix with value "-10*log10(Signal(1,1))" the Operator returns me the error: "ExpressionFilter::Preflight: State "log10" is inaccessible".
2) I really can not get how to configure Normalizer! In the case I would like NOT to have adaptation I should set Adaptation to 0 for each of the ExpressionFilter's output channel (only one, in this case, right?) and offsets and gains could have any value because they will not be taken into account. Am I right? But, if I would like to HAVE adaptation (in mean and variance both) WHICH and HOW MANY parameters should I set at the beginning of the experiment? Is the number of parameters equal to the number of control signals, e.g. only one in this case (the one that controls the y-direction movement)??
Thank you in advance,
Giulia