Difference between revisions of "Talk:Characterization of the NIA signal"
m (New page: == November 1st 2010 == After few tests we have done with the first version of the signal acquisition signal provided by Andrew Junker, and working underl Windows with USA settings, we ...) |
(→Matlab converter: new section) |
||
| Line 21: | Line 21: | ||
Another iportant poit would concern adaptation of the NIA to the user as an interface tool. Here, machine learning techniques can be used to learn the signal from the user when required to perform a task. | Another iportant poit would concern adaptation of the NIA to the user as an interface tool. Here, machine learning techniques can be used to learn the signal from the user when required to perform a task. | ||
| + | |||
| + | == Matlab converter == | ||
| + | |||
| + | Here i try to add the matlab function i've created to convert the .txt file from the Brainfinger software to a matlab dataset. A short help header is included. | ||
Revision as of 11:30, 2 November 2010
November 1st 2010
After few tests we have done with the first version of the signal acquisition signal provided by Andrew Junker, and working underl Windows with USA settings, we have noticed that:
o The available signal is at the moment a set of signals coming from (unknown) elaborations of the original raw signal that Andrew told us to come as a differential signal between left and right electrodes with respect to the central one
o There are differences in the signal due to the position of the electrodes; namely, at least two positions can be identified: just above the eyebrows and in the middle of the forehead; signal from eyes is of course stronger in the first position
o There is a strong influence of movements (eyes, but even more the jaw) on the signals, even the ones named with EEG names
Suggestions for the work to be done
We have two main lines that have to be investigated.
The first one is related to affective computing and can be pursued by performing test with NIA and, in parallel, the acquisition of bio-signals from the Pro-COMP (BVP, T, Resp...). here muscles are part of the interesting signal and it should be paid attention to avoid tasks where affective information can be confused with muscular activity required by the task.
The second one is related to BCI and EEG, and should be pursued by repeating typical BCI experiences (such as P300 and motor imagery) with a double instrumentation: NIA and EEG. Attention should be paid to avoid the use of muscles, as much as possible (in P300 it would be impossible).
Another iportant poit would concern adaptation of the NIA to the user as an interface tool. Here, machine learning techniques can be used to learn the signal from the user when required to perform a task.
Matlab converter
Here i try to add the matlab function i've created to convert the .txt file from the Brainfinger software to a matlab dataset. A short help header is included.
