Practical Neural Network Recipies in C++

13

Functional Link Networks

A simple transformation of the input data for a feedforward network

often dramatically speeds training. The input layer becomes much

larger, sometimes impractically so. However, the number of hidden

neurons can usually be greatly decreased, or eliminated entirely.

22a

Chapter 13

This chapter is largely based on material contained in [Pao, 1989].

That book is highly recommended not only as a source of additional

information about functional link networks, but also because it is an

excellent reference on neural network ...