RESEARCH OF MODEL FOR INCREASE LEARNING SPEED OF A NEURAL NETWORK IN THE LINUX OPERATING SYSTEM
Timur V. Jamgharyan · BULLETIN OF HIGH TECHNOLOGY · 2025
The paper presents the effect of changing the duration of the quantum of processor time on the efficiency of neural network training in the Linux environment. The speed of training of a convolutional neural network is considered with different parameters of the computing cluster. The obtained results demonstrate the possibilities of optimizing neural network training in a multitasking environment, increasing the overall computational efficiency. A detailed analysis of the relationship between resource allocation strategies and training speed is carried out, and recommendations for configuring Linux systems for working with neural networks are proposed.