Usally there is no fast rule but there are a number of
recomendations.
read these articles
1. Dimensionality and Sample size consideration in
pattern recognition practice
by. Anil K. Jain and B. Chandrasekaran
2. A problem with dimensionality
by G.V Trunk
as for your previous request about the mathematical
derivation check the attachment it might help.
Good luck
--- "[log in to unmask]" <[log in to unmask]> wrote:
> Hi, if the dimension of the features is around 100,
> is there a rule saying
> how large the training data set should be?
> Thanks
>
=====
The LORD is my shepherd; I shall not want. (psalm 23:1)
Araya, Seare Tesfamilchael
P.O. Box 308
Wits 2050
Johannesberg, South Africa
Tel:- 27-826385858
__________________________________
Do you Yahoo!?
Y! Messenger - Communicate in real time. Download now.
http://messenger.yahoo.com
|