this project aims to use the neural net framework and backpropagation
to learn to detect and recognize frontal, upright images of human faces.
Input images are taken from the at&t database, and can be passed to the learning application
 by a directory name.  The system will iterate recursively through all images given in the 
input directory and train its internal network based on the assumption that all images 
are of faces.

Negative training examples are necessary to prevent mis-applied learning of patterns by 
the network, and can be specified on the command line along with positive training instances 
and test(validation) instances..