Neural network training and its significance for the development of software engineering

Cozac Eugeniu

Currently, there is an increased interest in neural networks in various spheres of public life. However, the success of neural networks when used is not unconditional, since the main problem associated with the use of neural networks is their training. At the same time, the process of training neural networks is quite time – consuming, since it is necessary to choose both the task itself that the trained neural network should solve, and to prepare data sets for solving such a task-and the more complex the task that is assigned to the neural network, the larger and more disparate the data sets that need to be prepared. At the same time, when training neural networks, it is quite standard to have a situation in which the neural network copes with the task of training quite easily and quickly, and the learning process itself can be very slow. This is due both to the nature of neural networks, and to the fact that the solution of the problem of training neural networks and the choice of an algorithm from training depends on the direction of using neural networks. In the context of this article, the author considers certain aspects related to the training of neural networks and concludes that the training of neural networks is of great practical importance for the purposes of software engineering, since it allows solving more and more new tasks every year, and the importance of training neural networks is due to the ability of trained neural networks to accumulate large volumes of disparate data, the possibility of adaptation and improvement of neural networks with an increase in the performance of electronic computing equipment, as well as the ability of neural networks to adapt to constantly improving learning algorithms.

Keywords:machine learning, neural network training, deep learning, artificial intelligence, neural networks, neural network training algorithms.