bci2000.org BBS

Dear All

Hope you are fine and enjyoing your time . please help me in SWLDA Algorithm. i gave the input feature . i work for 7 training dataset and testing it on 8 datasets . i know the dtasets are small to train and test . but when i apply SVM on this it classify 7 our 8 characters while SWLDA classifying 5 characters . In this i used only one channel .

When i used 32 channel data ....SVM goes to 100 % results afetr tring on 7 datasets and testing on 8 datasets. While SWLDA give only 25% result. How SWLDA effiency decreases to much ??

I use the Matlab command .to implement SWLDA for my case . I am working on P300 signals.

please help me in this problem.


Regards

Hello , I don't know will you read this post but I hope will.
In fact, I have a problem to ask you. That is what kind of SVM parameters you set ? I am using libsvm .There are some parameters to set , such as -s -t -c -g and so on.
The feature extracted from P300 signal input SWLDA got adaptive performance , while in SVM got poor performance . I want to know where is the problem in my SVM method .

Thanks , sincerely
Sandra .