Intersectoral artificial intelligence technologies: search for and implementation of efficient solutions

Most studies in the area of artificial intelligence are associated with resolving the following contradiction. On the one hand, deep learning methods are universal and can be applied in various fields of research due to the generality of their basic mathematical and algorithmic ideas, software implementations, and the possibility of transfer of previously obtained learning results. On the other hand, the process of learning for solving a particular task requires specialized qualitatively labeled data, and high accuracy can be achieved by applying original algorithmic solutions and a proper tuning of hyperparameters. In the work of the Research Center for Artificial Intelligence of Innopolis University, this contradiction is resolved by creating an algorithmic core and corresponding hardware-software tools shared by the solutions of diverse intersectoral tasks. The scientific work of the Center is aimed at the creation of foundations sufficient for solving practical tasks. This paper covers the main results of scientific and practical works of the Center in 2022.