Hello everyone
It's been a while, I started to try to work with openvibe using g.Mobilab+ for capturing the EEG signals. I am working on linux (ubuntu) because the driver for this device only works with this OS.
I already worked with the different scenarios that you give as P300 and SSVEP paradigm. But for example in the case of P300 always i'm obtaining classification results of 60%. I have tried to analyze what could be causing the problem, and one of the things that makes me doubt, is the way I'm getting the data. Why whenever i'm capturing the data ..the device drift is activated..if i modified the number of samples per sent block change ..the same happens..
That's just one of the questions I have, this could be affecting my experiments?,
Thanks,
Luisa
Acquisition server and g.mobilab
Re: Acquisition server and g.mobilab
Hi Luisa,
If the drift correction is triggered too much, it can be a problem indeed.
If such high drift is detected, you should have in the console a warning message telling you how much the drift was corrected.
Is it the case ?
Maybe the g.Mobilab+ has a jitter that need more time to adjust itself... Try a higher drift tolerance value in the Acquisition Server preferences (from 2ms to 5ms).
Laurent
If the drift correction is triggered too much, it can be a problem indeed.
If such high drift is detected, you should have in the console a warning message telling you how much the drift was corrected.
Is it the case ?
Maybe the g.Mobilab+ has a jitter that need more time to adjust itself... Try a higher drift tolerance value in the Acquisition Server preferences (from 2ms to 5ms).
Laurent
Re: Acquisition server and g.mobilab
Hi Laurent,
Thanks for your answer,
Actually yes the data adquisition is triggering too much the drift correction ...the values of this drift are near to -3.42 ms to 3.66ms (values are constantly moving between this scale). Originally was using a correction of 2ms .. for that reason the triggering was active often. Now I try with 5ms and i think the behavior is more stable. In fact i dont understand very well what is the real meaning of this drift. Why is this happening this behavior with the driver sometimes has many samples and sometimes does not complete the package. Will you have any idea why is that?
Still have not tried to complete the entire experiment scenario, because i have one more question, this is about the number of samples per block.This value should be chosen according to what parameter?, for example for the P300 experiment is fine if I choose 32 samples per block, knowing that would mean in the case of gmobilab+ (32 * (1/256)) = 125ms. Should I choose a larger number of samples?
Thanks
Luisa
Thanks for your answer,
Actually yes the data adquisition is triggering too much the drift correction ...the values of this drift are near to -3.42 ms to 3.66ms (values are constantly moving between this scale). Originally was using a correction of 2ms .. for that reason the triggering was active often. Now I try with 5ms and i think the behavior is more stable. In fact i dont understand very well what is the real meaning of this drift. Why is this happening this behavior with the driver sometimes has many samples and sometimes does not complete the package. Will you have any idea why is that?
Still have not tried to complete the entire experiment scenario, because i have one more question, this is about the number of samples per block.This value should be chosen according to what parameter?, for example for the P300 experiment is fine if I choose 32 samples per block, knowing that would mean in the case of gmobilab+ (32 * (1/256)) = 125ms. Should I choose a larger number of samples?
Thanks
Luisa