There is a new gold rush going on, but this time gold means deep learning. Brand-new start-ups and traditional enterprises are looking toward Neural Networks (NN) and deep learning. As one of the most powerful programming languages for data scientists, R is sailing on this tide.

One of the most promising backend deep learning engines is TensorFlow from Google. Keras permits the user to work with TensorFlow and some other great engines, allowing fast experimentation. Simple codes and a wide variety of modern neural architectures are some of the strengths featured by Keras, compared to competing frameworks.

This chapter will demonstrate how to design and train deep learning models using the R interface for Keras, but not before going through several core concepts of deep learning and NN. These concepts are not only meant to help with models' understanding the models, but also to point out what should and shouldn't be done with them.

Here is what readers can expect from this chapter:

• Learn how NNs are used in our daily lives
• Get to know the biological inspiration for Artificial Neural Networks (ANNs)
• Get an introduced to the model's pieces: nodes, activations function, and layers
• Learn the most popular learning algorithms
• Meet TensorFlow and Keras
• Fit a deep learning classification model from scratch using Keras