A. V. Kornaev, E. P. Kornaeva, I. N. Stebakov, “Theoretical prerequisites for physically justified machine learning and its applications to fluid dynamics”, Dokl. RAN. Math. Inf. Proc. Upr., 2022, Volume 508,Pages <nobr>100

ADVANCED STUDIES IN ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

Theoretical prerequisites for physically justified machine learning and its applications to fluid dynamics

A. V. Kornaev^a, E. P. Kornaeva^b, I. N. Stebakov^c

^a Research Center of Artificial Intelligence, Innopolis University, Innopolis, Russia
^b Department of Intelligence Systems and Digital Technologies, Orel State University, Orel, Russia
^c Department of Mechatronics, Mechanics, and Robotics, Orel State University, Orel, Russia

Abstract: Some laws of physics postulate that a quantity in the physical process under study has to take its extremal value. In this work, a generalization of a law of this kind is proposed and an approach is described in which artificial neural networks are used to minimize the power of internal forces and to simulate hydrodynamic processes for various applications.

Keywords: physically justified machine learning, deep learning, image segmentation, variational problem, objective functional.

UDC: 004.8

Presented: A. L. Semenov
Received: 28.10.2022
Revised: 28.10.2022
Accepted: 01.11.2022

DOI: 10.31857/S2686954322070128