ABSTRACT
This chapter provides a very brief introduction to the modern and currently developing field of machine learning called Deep Learning. The chapter is focused on the use of artificial neural networks, where the depth of learning is given by the number of hidden layers. The principle is mentioned, which is characterized by the ability to search for combinations of text elements (vectors of attributes describing documents), on the basis of which a high classification efficiency can be achieved. The gradual reduction of the number of neurons in the higher layers leads to the formation of relevant but often very abstract attributes. Due to the simple demonstration of the operation of the mentioned classification method, a common neural feed-forward network trained using the backpropagation algorithm is used. The usage is demonstrated by R package applied on real data.
