general classifier configuration

Forum for discussion on different signal processing algorithms
Locked
Yueqing Li
Posts: 51
Joined: 08 May 2007, 16:41

general classifier configuration

Post by Yueqing Li » 08 Jan 2009, 13:06

Hi,

I just transferred from version 1 to version 2. I followed the P300 tutorial on the wiki and have some questions (they may be very simple).

I am learning how to use the P300 copy mode with version 2. In the training session, after I have loaded the InitialP3Session.prm, I found the classifier matrix is as the following:

1 4 1 1
1 6 1 1

1. I know it is only an initial classifier rather than a subject-specific classifier. But why channel 1 is chosen? Since I don't use channel 1 in my experiment, can I use some other channel numbers, rather than channel 1? For example, channel 12. Or it really doesn't matter in the training session. I use channel 2, 6, 11, 12, 13, 14, 15, 16.

2. After the offline analysis, I will get some features and will configure the subject-specific classifier. In the tutorial, it is said that "In the third column, enter 1 as output channel". So, will the 3rd column---output channel always be 1? Does that mean the system has only one output channel?

Thanks.

gschalk
Posts: 615
Joined: 28 Jan 2003, 12:37

P300

Post by gschalk » 10 Jan 2009, 14:25

Yueqing,

You will have to adapt the classifier matrix depending on which channels and time points show the best response. You will have determined these channels/time points using the OfflineAnalysis program or using the P300GUI. Both are described in the documentation. I also suggest you go through the P300 tutorial again. Most recently, we improved it and made it more clear.


Gerv

Yueqing Li
Posts: 51
Joined: 08 May 2007, 16:41

Post by Yueqing Li » 11 Jan 2009, 23:07

Dr. Schalk:

Thank you for your reply.

I think you misunderstood my question. I mean the classifier for the training session, rather than the subject specific classifier (which will be built from the data in the training session by P300GUI or OA). After loading the InitialP3Session.prm, should I make any changes of the classifier?

gschalk
Posts: 615
Joined: 28 Jan 2003, 12:37

Classifier ...

Post by gschalk » 12 Jan 2009, 07:12

Yueqing,

Thanks for clarifying. In the initial training session, there is no feedback of brain signals to the user. Thus, it does not matter what you put in the classifier matrix. In fact, you could even use the DummySignalProcessing module that does not do any online signal processing at all.

Gerv

Locked

Who is online

Users browsing this forum: No registered users and 5 guests