Research indexJournal article
PROCESS AUTOMATION USING MACHINE LEARNING
Frankov
Academic Visions28 Feb 2026Open access
NobleIDark:/48914/w1/20261104/2EE0A9A3DOI10.66556/2786-586x.52.frankov-o
PROCESS AUTOMATION USING MACHINE LEARNING
Frankov · Academic Visions · 2026
NobleIDark:/48914/w1/20261104/2EE0A9A3DOI10.66556/2786-586x.52.frankov-o
Abstract
The article explores modern approaches to process automation using machine learning and artificial intelligence technologies. It is established that the rapid development of digital technologies, the increase in data volumes, and the complication of production and information systems create the need to transition from traditional algorithmic methods of automation to intelligent adaptive systems. It is shown that classical automation systems, built on rigidly defined rules, have limited capabilities in conditions of dynamic changes in environmental parameters, while machine learning models provide the ability to adapt, analyze large amounts of information, predict technical conditions, and automatically optimize processes in real time. Modern directions in automation development are analyzed, among which special attention is paid to large language models, transformative architectures, deep and reinforcement learning methods, and intelligent autonomous agents. It is determined that large language models allow for multi-step process automation, support for digital services, and intelligent interaction with the user. It is proven that the combination of transformer models with reinforcement learning methods provides effective resource allocation, adaptive control of communication networks, optimization of production processes and automated decision-making. The paper considers practical aspects of using machine learning in predictive maintenance systems of equipment, where data analysis from sensor networks of the Internet of Things provides early detection of anomalies and prediction of emergency situations. Special attention is paid to the automation of CAD design, digital pipelines for processing 3D assets and autonomous laboratories, in which machine learning is used for automatic planning of experiments and optimization of technological parameters. The main problems of implementing automated systems based on machine learning are investigated, among which the lack of balanced data sets, the complexity of integrating intelligent models into the existing infrastructure, high computational costs, issues of reliability and security of automated solutions, as well as the need to ensure the explainability of models are identified. It has been established that the effective functioning of such systems requires the use of robust algorithms, control mechanisms and adaptive control under conditions of uncertainty. Thus, automation of processes using machine learning is one of the key areas of development of modern information and production technologies. The use of intelligent models ensures increased efficiency of system operation, reduction of operation time, optimization of resources and reduction of human errors. It has been determined that the prospects for further research are related to the development of autonomous adaptive systems, integration of artificial intelligence with digital twins, cloud platforms and the Internet of Things, as well as the improvement of explainable and robust machine learning models for critical applications.