A combination of various factors has made Artificial Neural Networks (ANNs) one of the most important machine learning techniques available today. These factors include the need to solve increasingly complex problems, the explosion of data, and the emergence of technologies, such as readily available cheap clusters, that provide the computing power necessary to design very complex algorithms.

In fact, this is the research area that is rapidly evolving and is responsible for most of the major advances claimed by leading-edge tech fields such as robotics, natural language processing, and self-driving cars. 

Looking into the structure of an ANN, its basic unit is a neuron. The real strength of the ANN lies in its ability to use the power of multiple neurons by organizing them in a layered architecture. An ANN creates a layered architecture by chaining neurons together in various layers. A signal passes through these layers and is processed in different ways in each of the layers until the final required output is generated. As we will see in this chapter, the hidden layers used by ANNs act as layers of abstraction, enabling deep learning, which is extensively used in realizing powerful applications such as Amazon's Alexa, Google's image search, and Google Photos.

This chapter first introduces the main concepts and components of a typical neural network. Then, it presents the various types of neural networks and explains the different kinds of activation functions used in these neural networks. Then, the backpropagation algorithm is discussed in detail, which is the most widely used algorithm for training a neural network. Next, the transfer learning technique is explained, which can be used to greatly simplify and partially automate the training of models. Finally, how to use deep learning to flag fraudulent documents is looked at by way of a real-world example application.

The following are the main concepts discussed in this chapter:

• Understanding ANNs
• The evolution of ANNs
• Training a neural network
• Tools and frameworks 
• Transfer learning
• Case study: using deep learning for fraud detection

Let's start by looking at the basics of ANNs.