Hi Everyone,
I am doing research on a comparison of neural networks compared to
SVM. Does anyone know of a recent review article or paper that does a
direct comparison of the two (or more techniques) where a large portion
of the database was withheld as a truly independent test of the
algorithms produced.
This subset would not be used to generate or select the final model,
but run after all model generation and selection is complete.
Hopefully the article would pinpoint the advantage of SVM with a
specific example.
Thanks ,
Marc Parham
|