UDC 004.85
IMPROVING THE EFFICIENCY OF NEURAL NETWORK TRAINING PROCESS BASED ON GENETIC ALGORITHM MULTITHREADED IMPLEMENTATION
D. A. Perepelkin, Dr. in technical sciences, Professor, CAD Department, Dean of Computer Engineering
Faculty, RSREU, Ryazan, Russia;
orcid.org/0000-0003-4775-5745, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. N. Saprykin, Ph.D. (in technical sciences), associate professor, CAD Department, RSREU, Ryazan,
Russia;
orcid.org/0000-0002-3882-1301, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
A. A. Tikhonov, student, RSREU, Ryazan, Russia;
orcid.org/0009-0006-7128-8795, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The problem of optimizing the training process in neural networks (NN) is considered. The aim of the
work is to develop and evaluate the performance of a high-performance algorithm for finding an optimal set
of weights for a neural network aimed at solving a specific type of problem. Due to high computational complexity of neural network training task, multithreaded implementation of a genetic algorithm is proposed,
focused on the efficient use of computational resources of multiprocessor systems. Comparative testing with
back propagation algorithm was conducted on neural networks of various scales and orientations. The article
shows that the proposed multithreaded modification of genetic algorithm provides a significant reduction
in training time and high stability of optimization process.
Key words: neural networks, genetic algorithm, evolutionary computations, multithreading