In this chapter, we've seen the power of neural networks with respect to recognizing digits from zero to nine in images. Although the coding of the digits was very small in 5 × 5 images, it was important to see the concept in practice. Neural networks are capable of learning from data, and provided that real-world representations can be transformed into data, it is reasonable to state that character recognition is a very good example of the application of pattern recognition. The application here can be extended to any type of characters, under the condition that the neural network be presented all the predefined characters.

The next chapter will explore all the content seen in this book so far to present the reader with some options for the optimization and improvement of the neural network application, concluding the outline designed for this book.