Practical Neural Network Recipies in C++

254

Chapter 16

Different network models impose different constraints

on

their training

data. Some recommendations are universal, though. Since restrictions

peculiar to a single network model are discussed thoroughly in the

chapters devoted to those models, they will be only briefly mentioned

here.

This chapter will focus primarily on aspects of data preparation

that have more widespread application.

General Considerations

Neural network practitioners must remember that they are not

competing with their networks; they are working together as a team.

Neural networks can sometimes produce such startlingly excellent

results that there is a tendenc ...