Generic Stream Reader/Writer Problems

[RacDar]

Hi,

We are trying to understand the functionality of the Generic Stream Reader and the Generic Stream Writer. Moreover, we cannot understand the connection between these blocks and the rest of the p300 Speller scenario. We would like to know what happens when writing into such a file and how they are passing the signal acquired from the EEG headset to the rest of the process blocks.

Thanks

[Haneen]

I am wondering too about the generic stream writer. Regarding the name of the file that we have to specify, will the generic stream writer create a file of type .ov by itself named by the specified name?

I'm getting this error:

[WARNING] At time 0.000 sec <Box algorithm::Generic stream writer> Compression flag not used yet, the file will be flagged uncompressed and stored as is
[WARNING] At time 0.000 sec <Box algorithm::Generic stream writer> Could not open file [record-[2013.02.25-01.03.40].ov]
[WARNING] Box algorithm <Generic stream writer> has been deactivated because initialization phase returned bad status

[jlegeny]

Hello,

the Generic Stream Writer is a simple file writing box. Its purpose is to store signal for future reading e.g.: by training scenarios or for offline processing.

Generic Stream Reader reads the data back. In the case of the P300 scenario, both boxes are present. This is done so one can choose whether he wishes to do a replay of a previously recorded session (by connecting the Generic Stream Reader to the P300 box) or to do an online session (by connecting the Acquisition Client box). In order to do an offline session the data has to be recorded first of course. You will also have to point the Generic Stream Reader to the right file.

The warning about compression can be safely ignored.

Cheers
Jozef

[Haneen]

Thanks for the reply Jozef!

1. What about the second error of not abling to open the file?
[WARNING] At time 0.000 sec <Box algorithm::Generic stream writer> Could not open file [record-[2013.02.25-01.03.40].ov]

I guess I'm writing something wrong in the "File name" option of the Generic Reader/Writer! Can you please give an example of what should be written in the "File name" ? I have tried to specify a path for the generic writer in order to create the file in it but that is not working!

2. If I want to make an online session for the [P300-speller-1-acquisition], is that what you meant?


Looking forward you reply!
P.S: Sorry for posting my questions here but I didn't want to open a new topic while the topic is the same!

[tonio112003]

Hi!

I don´t know why but, when the Train Classifier ends, doesn´t generates the .cfg data

Can anyone help me please?

[jlegeny]

@Haneen:

you can precise the file to write in the generic file writer. In order for the file to be created the path must point to an existing directory and the filename must be supported by the operating system (for example on Windows you can not use some characters such as :,( or ).

The file will be created only when there is any data incoming, i.e., you need the acquisition server to be running and sending data to openvibe.

Stream reader simply reads any specified file, it should have as a parameter the exact location of the file you wish to read. You can select it by clicking the button at the right of the path field.

Your scenario for the online session is correct. I would, however, recommend using the XDAWN versions of the P300 scenarios.

@tonio112003:

I don´t know why but, when the Train Classifier ends, does not generates the .cfg data

is there any error message displayed? Can you see the output of the K-fold test from the classifier trainer?

[tonio112003]

jlegeny:

I made an adquisition training for 16 letters(trials) and 15 repetitions each. The .ov document is 48.8 MB big. I don´t know if it is too big?

I understand that, in the classifier training, is offline but I see the software makes a repetition of the whole training and then does not anything happen, it does not show an error message.

It should show a result, right?

I read that the number of partitions makes it slower so I put it in 2 partitions but it does not work anyway.

[Haneen]

Thanks Jozef for your reply!

I have shifted to xDawn version of the P300. It's easier to deal with and clearer.

Now I am done with signal monitoring, acquisition , xDawn training , classifier training SUCCESSFULLY!

But the online mode is not working properly. What I did in this scenario:
1.Disconnected the target letter generator.
2.Direct the xDawn filter to the cfg file produced before.
3. Direct the 12 classifier processors to the cfg file produced before.


I am getting tens of this warning:
[WARNING] <Algorithm::LDA classifier - GPL> Feature vector size 57 and hyperplane parameter size 229 does not match

So:
1. What does this warning mean that warning mean?

2. I am not getting a Result value. I mean when I want the program to guess a letter, it moves to the second letter and it doesn't guess. A corresponding warning that I am getting:
[WARNING] <Algorithm::LDA classifier - GPL> Feature vector size 57 and hyperplane parameter size 229 does not match


Any suggestion?!

[tonio112003]

I have exactly the same problem. I do not use xDawn but after classifying, when I want to start the Online part, I have the message

[WARNING] <Algorithm::LDA classifier - GPL> Feature vector size 384 and hyperplane parameter size 307 does not match
[WARNING] At time 13.531 sec <Box algorithm::Voting Classifier> Input matrix should have a single value
[WARNING] Box algorithm <Voting Classifier> has been deactivated because process phase returned bad status

I would like to know why does it happens too.