Practical Neural Network Recipies in C++

116 Chapter 6

method having probability one of finding a global minimum among

local minima. There is potential here.

When to Use a Multiple-Layer Feedforward Network

This model has a well-deserved reputation for being the neural

network. It is a universal function approximator. We can, theoretical-

ly at least, teach anything learnable to this network. Thus, it is

reasonable to consider this network for any problem.

Another advantage of the multiple-layer feedforward network

is that its execution speed is among the fastest of all models currently

in use. Many models, none of which is presented in this book, use

iteration to converge to a decision ...