Signal concatenation

Concerning processing components: filters, file load/save, visualizations, communication ...
Post Reply
catha
Posts: 4
Joined: Mon Feb 18, 2013 1:03 pm

Signal concatenation

Post by catha »

Hey,

I have some problems with the box ‘Signal Concatenation’. I am trying to combine two datasets which share the same characteristics to improve the classification accuracy. I understood the box settings as follow:

Time out before assuming end-of-file (in sec) – I thought this equals the seconds which are cut out from the dataset before or after the trigger which I specify in the next setting field. But this seams not the case because the overall size of the dataset does not change if I enter 3 or 20 seconds. So does somebody know what exactly happens here?

End-of-file stimulation for input 1 – I understood that I specify here the trigger which equals the end of the file. Because this exists at least twice once for each input it should be that the end of two datasets doesn’t have to be the same.

As an example the motor imagery paradigm has at the end a ExperimentStop and 5 seconds later a StimuationId_Train trigger. Now I want to concatenate two datasets which have both these triggers. After concatenation the first one should not have the StimulationId_Train trigger, but the second one should still have this. I thought the best solution would be set Time out before assuming end-of-file (in sec) to 3, End-of-file stimulation for input 1 to ExperimentStop and , End-of-file stimulation for input 2 to StimulationId_Train. But this doesn’t work. After some attempts I figured that I always lose one trigger in both dataset or lose no one at all. This means if both datasets still have both trigger the classifier will stop to early, after the first. On the other hand if the trigger is deleted for both the classifier will never stop.

So is there a possibility that I retain one more trigger in the second dataset as in the first so that this one could be used to stop the classifier?

I hope you could help me to understand this box and to solve this problem.
Catharina

jlegeny
Posts: 239
Joined: Tue Nov 02, 2010 8:51 am
Location: Mensia Technologies Paris FR
Contact:

Re: Signal concatenation

Post by jlegeny »

Hello catha,

the Time out setting actually triggers the end of file after all of the data have been received not after. Its name can be read as "wait for N seconds before giving up and considering the file finished". This is done because EEG data is not necessarily continuous and you do not know if there is any marker at the end of the file.

In your case what you could try is setting the first file's end of file trigger to StimuationId_Train and the second's to something which is not there. Then just set the timeout to 5 seconds or so. First file will end on the Train trigger and this should be removed, the second will timeout and all of the stimulations will be inside.

Hope this helps
Jozef

catha
Posts: 4
Joined: Mon Feb 18, 2013 1:03 pm

Re: Signal concatenation

Post by catha »

Hey Jozef,

thanks for your fast answer, now i understand what time out means.
Unfortunatly your suggestion regarding the trigger was not really working, I still had all in the file. But I found the box 'Stimulation Filter' so I could easily delete one trigger in the first dataset.
Anyway thank you

catha

Post Reply