Registration Number
0224U031750,
0122U002026
, R & D reports
Title
Computational algorithms and optimization for artificial intelligence, medicine and defense
popup.stage_title
Head
Liashko Serhii I., Доктор фізико-математичних наук
Registration Date
11-06-2024
Organization
Taras Shevchenko National University of Kyiv
popup.description2
Improved algorithms for solving inverse problems of substance transfer in a porous medium have been developed. This approach provides an opportunity to develop optimized micro-irrigation systems and to make optimal decisions regarding the purification of groundwater from pollution, which is relevant when solving environmental problems.
New methods for evaluating machine learning models for forecasting the epidemiological curves of COVID-19 have been developed. A new non-parametric random error test and its application to compare five models for forecasting the epidemic disease of COVID-19 have also been developed. For the first time, this test is proven to have a guaranteed level of significance due to the use of a precise confidence interval, which guarantees the effectiveness of predicting the epidemiological curves of COVID-19.
New decentralized algorithms for variational inequalities and game problems are developed and researched. Such algorithms have a dual purpose and can bring benefits for strengthening the national security and defense of Ukraine in the field of information technologies of artificial intelligence, machine learning models and neural networks.
New homogenization methods for variational inequalities and equations are investigated and justified. Such inequalities and equations describe all kinds of processes. Calculation of homogenized trajectories of such processes makes it possible to simulate real processes cleared of various noises and disturbances. In particular, homogenized trajectories of equations guarantee the most comfortable use of microneedle arrays, which are gaining popularity as a new injection method in modern medicine. Accordingly, variational inequalities are, for example, used to implement artificial intelligence algorithms, machine learning models, and neural networks. Homogenization of such inequalities determines the most optimal trajectories of such algorithms.
Product Description
popup.authors
Veklych Olena A.
Kashpur Olena F.
Kliushyn Dmytro A.
Sirenko Ihor P.
Hennadii V. Sandrakov
Semenov Volodymyr V.
Stelia Оleh B.
popup.nrat_date
2024-06-11
