Abstract
The article presents the basic concept of competitive learning in neural networks. Provides the main machine learning learning models and applications. The analysis of the advantages and disad-vantages of these models is carried out. The geometric interpretation of competitive learning is presented in terms ofmathematical formulas, as well as the behavior of neurons in this model. Neural systems are described as a powerful tool and driver in the field of modern Internet technologies, in data science and big (meta) data.