#8 serialLDA comment on transform

[virginia estellers]

Hi,

I've been reading on LDA transform and would like to know if the .apply() method from serialLDA it's a dependant or independant-LDA i.e you get a linear transform that in general works ok for all the classes as the eigenvectors of inv(Sw)*Sb or there is a linear transform for each class (corresponding to the eigenvectors of inv(Sw)*Sb_class1) and when you apply it to the data in fact each of these transforms is applied and its result compared to the means of the transformed classes and once yoy decide for a class you also decide the transform to apply to your data.

In the serialLDA.h file, line 166 you can find comments that induce me to think of the second option, but the code on the serialLDA_template to me looks as the first kind of general transform. Am I looking in the wrong place for the code or the comments do not apply?

I'd be much grateful for an answer.
Thanks in advance and best regards

Virginia