In Chapter 6, we discussed some aspects of functioning and teaching a single-layer neural network built from nonlinear elements. Let us continue the analysis now, showing how multilayer nonlinear networks work. They present more significant and interesting possibilities, as we saw from working with the Example 06 program.
You now know how to build multilayer networks from nonlinear neurons and how a nonlinear neuron can be taught, for example, by modifying Example 06. However, you have not yet encountered a basic problem of teaching such multilayer neural networks built from nonlinear neurons: the so-called hidden layers (Figure 7.1). What is the impact of this problem?