Artificial neural networks one of the basic directions of the modern theory of an artificial intelligence. In the first part of work biological preconditions of creation of the theory of artificial neural networks are stated. It is defined formal neuron. Single-layered and multilayered neural networks are considered. The basic theorem for approximation of functions by means of multilayered neural networks is resulted and the base algorithm of training of neural networks is described. General characteristics of the relaxing and self-organizing neural networks are given. The second part of work is devoted to application of artificial neural networks in various areas. The example of movement of the mobile robot on the basis of the rangefinder’s data is resulted. Recognition during movement of the mobile robot of a local relative attributes is considered, and application self-organizing maps at distribution of financings of institute is also described.

Theoretical and applied problems of mechanics