I am struggulling with OpenVine for 3 weeks. I will use P300 magic for visual event related potantial BCI application. Images and backround can be changed easily for request.
1. a. In acquistion.xml data recorded in signal folder. Here subject ID must be exist in data file name. How can I add ID to filename?
b. Acquistion.xml trial and session number in gui are not same as application.
c. Why all images do not stimuli in aquiation.xml ? Only some of them flashes. In acquistion, I think data from all images must be recorded.
d. I tried to convert .ov file in to .csv or .edf as follows, generic file reader and csv/edf writer but it failed.
2. a. In train-classifier.xml file, defaultly generic stream reader reads signal/bcı-p300-signal.ov. Why only one file is used? For example, how can I data from 18 subjects?
b. How can MATLAB classifer be applied in this part?
Please help me.
Apdullah YAYIK
my email: apdullahyayik@gmail.com
P300 Magic
-
jtlindgren
- Posts: 775
- Joined: Tue Dec 04, 2012 3:53 pm
- Location: INRIA Rennes, FRANCE
Re: P300 Magic
Hello Apdullah,
I can only recommend you to examine carefully the components of the used scenarios and read the associated web documentation about boxes and scenarios. To help a little,
a) you can add subject name manually by editing the Generic Stream Writer box config. Be careful to load the same file later!
b) not sure I understand this problem, can you give details?
c) the order is randomized. Perhaps some images are missed by chance? You can modify share/openvibe/scenarios/bci-examples/p300-magic-card/p300-magic-card-stimulations-targets.lua to change what stimulations are created.
d) these components should in principle work, but you don't give enough information to find out whats the problem. Compare your box settings to those of scenarios having boxes that read or write. Do you get some concrete errors?
Currently OpenViBE is working best with data from one subject -- the folklore claims that at least with popular classifiers like LDA, adding data from different subjects does not improve performance (e.g. due to usual differences in cortical activity, electrode positions, subject moods, noise conditions, sunspots, etc). More elaborate methods including e.g. data alignment might exist, but are not implemented in openvibe. If you want several subjects data to build a model, you need to combine all your data to a single file.
Cheers,
Jussi
I can only recommend you to examine carefully the components of the used scenarios and read the associated web documentation about boxes and scenarios. To help a little,
a) you can add subject name manually by editing the Generic Stream Writer box config. Be careful to load the same file later!
b) not sure I understand this problem, can you give details?
c) the order is randomized. Perhaps some images are missed by chance? You can modify share/openvibe/scenarios/bci-examples/p300-magic-card/p300-magic-card-stimulations-targets.lua to change what stimulations are created.
d) these components should in principle work, but you don't give enough information to find out whats the problem. Compare your box settings to those of scenarios having boxes that read or write. Do you get some concrete errors?
Currently OpenViBE is working best with data from one subject -- the folklore claims that at least with popular classifiers like LDA, adding data from different subjects does not improve performance (e.g. due to usual differences in cortical activity, electrode positions, subject moods, noise conditions, sunspots, etc). More elaborate methods including e.g. data alignment might exist, but are not implemented in openvibe. If you want several subjects data to build a model, you need to combine all your data to a single file.
Cheers,
Jussi
-
apdullahyayik
- Posts: 9
- Joined: Sat Jul 11, 2015 10:02 am
Re: P300 Magic
For 1 c.
My card_base is OVTK_StimulationId_Label_00,
Random tagets refer from lua script below;
box:send_stimulation(1, card_base + math.random(card_count) - 1, box:get_current_time() + delay, 0)
How can I avoid randomness and ordered targets so I tried this but not work,
box:send_stimulation(1, card_base + stimulation, box:get_current_time() + delay, 0)
Thank for helping..
Apdullah YAYIK
My card_base is OVTK_StimulationId_Label_00,
Random tagets refer from lua script below;
box:send_stimulation(1, card_base + math.random(card_count) - 1, box:get_current_time() + delay, 0)
How can I avoid randomness and ordered targets so I tried this but not work,
box:send_stimulation(1, card_base + stimulation, box:get_current_time() + delay, 0)
Thank for helping..
Apdullah YAYIK
-
apdullahyayik
- Posts: 9
- Joined: Sat Jul 11, 2015 10:02 am
Re: P300 Magic
Now, I have handled problem above...
But another one has emerged.
I have doubts about sycronization of aquistion server and simulations.
When I played designer eeg recording is starting and .0v file or .edf file are saved. But simulation starts when I push "a" button.
In .edf file I cant see when I pushed "a" button.
Can help me please.
But another one has emerged.
I have doubts about sycronization of aquistion server and simulations.
When I played designer eeg recording is starting and .0v file or .edf file are saved. But simulation starts when I push "a" button.
In .edf file I cant see when I pushed "a" button.
Can help me please.