Interdepartmental Research Center for Artificial Intelligence

Purpose

Combining the efforts of specialists in the field of medical, biological, information, technical and physical-mathematical sciences to create advanced and competitive technologies of artificial intelligence and creating conditions for the implementation of research projects in the field of artificial intelligence, implemented both within the framework of state research programs and with the involvement of non-state investments.

Directions

• analysis of biomedical data and images, image search and pattern recognition in super large databases of medical imaging
• intelligent systems for analyzing text information, thematic search engines in indexed databases
• methods and algorithms of automatic analysis, recognition and speech essitis in human-machy systems
• application of the apparatus of artificial neural networks for modeling cognitive processes and adaptive behavior in robotic control systems
• cognitive technologies for deepening ideas about brain plasticity in normal and pathology
• cellular technologies, stem cells and interface-brain-computer system to restore brain function after injuries and strokes
• nanoparticles and Heterocyclic compounds in experimental and cellular technologies in brain tumors

Results obtained

• The algorithm of prediction of the field of binding of home meter protein complexes based on a neural deep learning network, which for 53% of the tested complexes correctly identified most of the interface elements, and providing a significant reduction in the number of possible orientations of protein structures relative to each other when modeling complexes.
• Algorithm for modeling the structures of dimer protein complexes based on the contact matrix, which allowed to obtain correct three-dimensional models on the test set of protein complexes for 94% of heterodimeters and 96% of the geterodimers.
• Using computer screening and molecular modeling technologies, an analysis of 213.5 million chemical compounds was carried out, as a result of which 5 compounds capable of blocking the SARS-CoV-2 coronavirus site, critically important for its life cycle, have been identified. It is shown that the detected compounds form promising basic structures for the creation of effective drugs for the treatment of COVID-19.
• Methods, algorithms and software based on (Deep Learning) and (Convolution Neural Networks) for solving relevant application problems of analysis and recognition of digital images. Algorithms have been built and high-performance software tools with massive parallelism (more than 3000 cores) have been built to analyze full-slide (seltachios of several gigapixels) digital histological images of cancer patients tissues for an objective automated assessment of the degree of aggressiveness of tumors.
• Experimental sample of the information system of intelligent processing and storage of large volumes of medical and genetic data on lung diseases based on high-performance systems.
• Intelly software complex for forecasting risks of developing bone sarcome recurrives in children and young adults, taking into account the expression of molecular markers in tumor tissue.
• A prototype of an intelligent analytical system for processing and analyzing time-to-gene clinical and biomedical data in order to improve the diagnosis of complex diseases.
• Software complex control equipment for critical size control systems based on machine vision systems. The volume of production (innovation) of newly mastered (new) products in 2020 in physical terms at the enterprise "KBTEM-OMO" OJSC "Planar" amounted to 5 complexes, in monetary terms - 375 thousand. $. United States.
• The subsystem of Internet monitoring of the intelligent web-system for searching and processing texts in English and Russian by the method of sensing for the method of sensing of the intelligently coding method for information support of the solution of management, design-search and expert tasks in the field of remote sensing of the Earth.
• For the first time 58 text and speech services www.corpus.by were presented in the European infrastructure for access to language electronic resources and technologies CLARIN. Now they are available at the address vlo.clarin.eu. This made it possible to increase the visibility of the experience of domestic scientists in the field of processing Belarusian texts and speech, as well as the opportunity to apply for projects to international funds together with large European consortia and teams.
• The system of the audio guide for Internet, Android and Ios platforms on the historical and cultural exhibits of the Brest Archaeological Museum krokam.by/biarescie.
• Algorithms and Experimental Real-time Text and Speech Processing Software www.corpus.by (including 20 services with API).
• Automated systems for issuing current information for the population of the City Gynecological Hospital, the 4-th city clinical hospital named after M.Y. Savchenko, the Committee on Health of the Minsk City Executive Committee using speech synthesis.
• Mobile Internet system to assess the soda characteristics of normal and pathological speech.
• The system of audio guides for the Internet, Android and Ios platforms on the historical and cultural exhibits of the Brest Archaeological Museum (4th, final stage). krokam.by/biarescie
• Prototype of control system for non-industrial microrobots of a new generation, based on such principles as:
  • application of hybrid systems that combine analog and digital computing environments;
  • application of highly specialized filter modules, PID regulators, inclinomemeters, gyroscopes and others to solve typical microrobot control problems;
  • application of nonlinear dynamic systems, including artificial neural networks and chaotic generators as a computing environment;
  • application of the most universal computational environments for recognition and forecasting problems in order to solve the largest number of problems by each of the hardware modules;
  • focus on hardware implementation;
  • focus on energy efficiency of calculations.
• Micromodule of operational recognition, selection and compression of video information on board small spacecraft.
• An experimental software complex for the maintenance of objects and automatic detection of moving objects observed by the onboard camera of the unmanned aerial vehicle has been developed. The developed software complex is designed to create a subsystem for multi-purpose video image processing obtained by a video camera of an unmanned aerial vehicle.
• The software complex of delivery and processing of textual scientific and technical information for the formation of an electronic data bank of new technologies and materials that ensure the creation of rocket and space technology and promising products of other industries that are competitive in the world market. The created software is transferred for use to the state scientific institution "Institute of Heat and Mass Exchange Named A.V. Lykova of the National Academy of Sciences of Belarus" based on the results of the program of the Union State "Technologia-SG".
• The software information complex that implements an automated technology for the preparation of a regional forecast of the oil and gas of the territory on the basis of a comprehensive analysis of Earth remote sensing data from space and geological and geophysical information. The developed software is designed for space mapping of oil-prospective objects and assessment of oil and gas bases of sediment basins based on specialized software processing of remote sensing data and a comprehensive interpretation of space and geological and geophysical information. The complex was put into operation in the open joint-stock company "Belorkhimprom", the state scientific institution "Institute of Nature Management of the NAS of Belarus" and the Belarusian State University for the purpose of carrying out works related to the search for hydrocarbon deposits, as well as for the training of students and undergraduates.
• Imitation model of a typical logistics warehouse, which is served by transport robots. Each of these robots is equipped with a manipulator with a grip, equipped with vacuum suction, and a developed technical vision system designed not only for safe motion, but also for determining the spatial position of the desired product on the rack.
• A universal IoT module designed for remote control of mobile platforms has been developed and explored. This module is a complete and self-sufficient device that has a number of modern sensor devices that allow you to control mobile objects in space, as well as wirelessly communicates with a remote server, which significantly increases its computing power and memory capacity. This module can be used as a transport rental element (with further integration into the payment system), or remote control of transport trouts.
• The software of the automatic control system of the austerary culturer has been developed, including video sequence processing modules with the allocation of the center of rows, making a decision on the displacement of the mobile part of the cultivator to the desired direction, as well as an ergonomic interface designed to work on a touch monitor, which is installed in the cockpit of the leading tractor. This system was successfully responsive and tested on the experimental site of the real field of sugar beet.
• Software and hardware designed to determine in real time to determine the external loads on potato tubes in the process of cleaning and post-absorpation and in case of exceeding the permissible values of the information operator of the system. The developed system is placed in the body "artificial potatoes", which will subsequently be used in the field to determine the critical loads obtained during the process of harvesting and transporting potatoes from the field to the warehouse