Just as there are many different types of learning and approaches to human learning, so we can say about the machines as well. To ensure that a machine will be able to learn from experience, it is important to define the best available methodologies depending on the specific job requirements. This often means choosing techniques that work for the present case and evaluating them from time to time, to determine if we need to try something new.

 We have seen the basics of neural networks in Chapter 1, Neural Network and Artificial Intelligence Concepts, and also two simple implementations using R. In this chapter, we will deal with the learning process, that is how to train, test, and deploy a neural network machine learning model. The training phase is used for learning, to fit the parameters of the neural networks. The testing phase is used to assess the performance of fully-trained neural networks. Finally, in the deployment phase, actual data is passed through the model to get the prediction.

The following is the list of concepts covered in this chapter:

• Learning process
• Supervised learning
• Unsupervised learning
• Training, testing, and deploying a model
• Evaluation metrics-error measurement and fine tuning; measuring accuracy of a model
• Supervised learning model using neural networks
• Unsupervised learning model using neural networks
• Backpropagation

By the end of the chapter, we will understand the basic concepts of the learning process and how to implement it in the R environment. We will discover different types of algorithms to implement a neural network. We will learn how to train, test, and deploy a model. We will know how to perform a correct valuation procedure.