Hi all,
in the application,i want to use the C3 and C4 channal to control the cursor up and down. If i image right hand move movement (according to the C3 channal),the ball will up; and if i image left hand move movement(according to the C4 channal), the ball will down. There is only 1D control,use C3\C4 to control up and down, i have several problems.
1 in the linear classifier, is the output channal should be set to 1(or 2) to control the ball up and down? and is there any directions in BCI2000 to define this problem? Or BCI2000 has fix the output channal number in classifier correspond to 1(horizon),2(vertical) 3(depth), and this is apply to all applications?
2 in order to control the cursor up and down, what should i set in the weight parameter in the linear classifier? i have set it like this
input channal input element output channal weight
C3 12Hz 1 -1
C4 12Hz 1 1
and i have puzzle that is it the final output is positive(+) and the cursor will go up and negative(-) the cursor will go down? is there any directions in BCI2000?
Thanks a lot!
FeedbackDemo application problems
-
mellinger
- Posts: 1341
- Joined: 12 Feb 2003, 11:06
No, these directions only apply to the CursorTask application module. In the FeedbackDemo module, output channel 1 is used for vertical movement.Or BCI2000 has fix the output channal number in classifier correspond to 1(horizon),2(vertical) 3(depth), and this is apply to all applications?
Yes, a positive control signal will move the cursor up, and a negative control signal will move it down. This applies to both the CursorTask and the FeedbackDemo application module.and i have puzzle that is it the final output is positive(+) and the cursor will go up and negative(-) the cursor will go down?
Regards,
Juergen
-
axone.he
- Posts: 14
- Joined: 27 Nov 2010, 09:11
Re: FeedbackDemo application problems
Juergen, it is great to konw,thank you.
I have done my test.
And the highest accuracy is 80% and the lowest is 60% for 8 times tests, each test contains 20 trials. The result is for a beginner.
1 I want to konw whether the result is convinced that the configuration is right?
2 I use three modules, neuroscan for acquistion, ARsignalprocessing for signal processing, FeedbackDemo for application, I have set the AR order for 16,and the windowlength 0.5s,bufferlength 12s corresponding to the feedbackduration 6s. Except for the parameter in linearclassifier,whether these configurations is suitable? And do you have some advice about the configurations to make the system work better.
Thanks a lot
Best regards
I have done my test.
And the highest accuracy is 80% and the lowest is 60% for 8 times tests, each test contains 20 trials. The result is for a beginner.
1 I want to konw whether the result is convinced that the configuration is right?
2 I use three modules, neuroscan for acquistion, ARsignalprocessing for signal processing, FeedbackDemo for application, I have set the AR order for 16,and the windowlength 0.5s,bufferlength 12s corresponding to the feedbackduration 6s. Except for the parameter in linearclassifier,whether these configurations is suitable? And do you have some advice about the configurations to make the system work better.
Thanks a lot
Best regards
-
mellinger
- Posts: 1341
- Joined: 12 Feb 2003, 11:06
Re: FeedbackDemo application problems
Yes, your results are realistic for an untrained subject.1 I want to konw whether the result is convinced that the configuration is right?
Yes, the configuration is suitable. However, a feedback duration of 6s is quite large. Most subjects will not be able to sustain an event-related desynchronization for so long. You might try 3s instead.2
-
axone.he
- Posts: 14
- Joined: 27 Nov 2010, 09:11
Re: FeedbackDemo application problems
Thanks a lot for you help, i would try to make the application better.
Best regards
Best regards
Who is online
Users browsing this forum: No registered users and 0 guests
