Get the entire book!

Multi layer neural networks are necessary to allow more advanced patterns to be recognized. Multilayer networks are particularly necessary for non-linearly separable problems such as XOR. This chapter showed you how to use a neural network class to process the XOR problem.

We begin by looking at a open source editor for neural networks. JOONE includes a graphical editor that allows you visually create a neural network. This editor works fine for prototyping networks but neural networks created by the editor cannot easily be incorporated into an actual Java program. To incorporate JOONE neural networks into an actual Java program the JOONE Engine must be used. In this book we will develop our own class to evaluate neural networks. The internals of this class will be explained further in Chapter 5.

In this chapter we saw the process by which the neural network is trained, and ultimately used. An array if input signals is presented to the neural network, and a corresponding array of output signals is processed. Error can be calculated based on how different the actual output was from the anticipated output. In the next chapter you will be introduced to some of the methods by which neural networks can be trained. Then Chapter 5 will introduce you to the backpropagation method that the neural network class presented in this chapter uses to train.