Chapter 4: Deep Learning with Neural Networks

This chapter is an introduction to neural networks with Keras. If you have already worked with MNIST or CIFAR-10 image classification datasets, feel free to skip it. But if you have never trained a neural network, this chapter might have some surprises in store for you.

This chapter is quite practical, to give you very quickly something to play with, and we will skip as much theory as reasonably possible and learn how to recognize handwritten numbers (composed of one single digit) with high precision. The theory behind what we do here, and more, will be covered in the next chapter.

We will cover the following topics:

• Machine learning
• Neural networks and their parameters
• Convolutional neural networks
• Keras, a deep learning framework
• The MNIST dataset
• How to build and train a neural network
• The CIFAR-10 dataset