Hello,
I have a question. Neural networs, SVMs etc. all have this term "bias".
All I know about its purpose is to position the hyperplane when the data is not
at the origin i.e. whe the solution does not go through the origin.
In the event one wants to get rid of the bias, one has to
centre the data i.e. zero mean and unit variance which is actually
normalising the data.
However, I see that even with the bias term people normalise their data.
Why so? If someone can clarify the whole idea of the bias term and why one
needs to normalise if one is using the bias, I would really appreciate it.
Thank You.
Sincerely,
Monika Ray
***********************************************************************
The sweetest songs are those that tell of our saddest thought...
Tulane Electrical Engineering and Computer Science, New Orleans, LA, USA
***********************************************************************
|