Intelligent Information
Technologies for Industry

Sergey Petrenko

D.Sc. (Tech.), Sirius University of Science and Technology, Russia

Technologies to counter the new quantum cyber threat

The relevance of ensuring the quantum resilience of national blockchain ecosystems and platforms in the Russian Federation's digital economy is explained by the emergence of a new quantum threat, growing security requirements for critical information infrastructure in the Russian Federation, the growth of the structure and behavior of blockchain ecosystems and platforms, and the insufficiency (inability) of known technologies (models, methods, and means) for ensuring information security and cyber resilience to detect, neutralize, and prevent the aforementioned threat. The achievements of IBM and a number of other high-tech quantum computer manufacturers provide strong evidence of the reality of the so-called “quantum threat.” For this reason, a number of technologically advanced countries around the world have already begun preparing to counter future quantum cyberattacks.

The goal of the research is to create promising world-class technology (mathematical models, methods, and tools) to ensure the quantum resilience of leading national blockchain ecosystems and platforms of the Russian Federation's digital economy (Enterprise Ethereum Alliance, Waves Enterprise (Waves, Vostok), Hyperledger Fabric (Linux, IBM), Corda Enterprise, Masterchain, Microsoft Azure Blockchain, etc.) in the face of previously unknown cyberattacks by malicious attackers using quantum computers, which, unlike existing technologies, will make it possible to prevent the aforementioned blockchains from suffering significant or catastrophic consequences. This includes developing new methods and algorithms for analyzing and parametrically synthesizing quantum-resistant blockchain ecosystems and platforms for the Russian digital economy, as well as corresponding methodologies and prototypes for software complexes to ensure the quantum resistance of these blockchains in the face of new quantum security threats. It is important to note that, unlike existing technologies, the proposed technology makes it possible to prevent the aforementioned blockchains from suffering significant or catastrophic consequences in the event of previously unknown attacks by malicious actors using quantum computers.

Speakes BIO

Head of the research group at the Scientific Center for Information Technology and Artificial Intelligence. Head of the State Scientific School “Mathematical Support for Critically Important Objects (CIO) of National Infrastructure” (since 2011). Expert in the Information Security Section of the Scientific Council under the Security Council of the Russian Federation (since 2013). Professor at several state universities, in 2024 he was ranked among the top 10 leading Russian scientists in computer science and computer security.

Doctor of Technical Sciences (2011), specializations: 20.02.12 – System Analysis, Mathematical Modeling, Computer Technologies (Technical Sciences); 20.02.27 – Information Warfare (Technical Sciences). Scientific editor of the journal "Inside. Information Protection". Author and co-author of 17 monographs and more than 550 articles on computer security issues. He has created a new direction in the field of artificial intelligence for protecting cyber-physical systems from malicious computer attacks.

Research interests: computer science, technological sovereignty, data economy, digital economy, cross-cutting information technologies, quantum technologies, artificial intelligence, blockchain ecosystems and platforms, quantum informatics, quantum computing, quantum algorithms, technology (models, methods, and tools) for ensuring the quantum resilience of digital platforms and blockchain.

Liudmila Zhilyakova

D.Sc. (Physics and Mathematics), Institute of Control Sciences, Russian Academy of Sciences, Russia.

The Problem of Time Encoding in Biologically Inspired Intelligent Systems

The encoding of temporal information is a basic function of biological systems that underlies perception, learning, decision-making, and behavioral synchronization. Despite an extensive array of empirical data and a variety of theoretical models, there is still no unified concept of temporal encoding. The report provides an overview of publications dedicated to research on the memorization, representation, and reproduction of time intervals by living organisms, as well as the modeling of time encoding.

The modern understanding of neural correlates of time encoding is considered. The evolution of approaches and models is traced: from classical models of the internal chronometer to more complex network, population, and Bayesian concepts. Key trends and paradigm shifts in modeling memorization and prediction of time are outlined. The key properties of time encoding observed in most vertebrate species are indicated.

Two fundamental models are described in detail: the internal clock model and the Scalar Expectancy Theory. Their significance for control theory, artificial intelligence, and robotics is justified.

Speakers BIO

Head of the Laboratory of Network Models in Neuroinformatics and Multi-Agent Systems at the Trapeznikov Institute of Control Sciences, Russian Academy of Sciences, Moscow, Russia; D.Sc. (Physics and Mathematics). She graduated from Rostov State University (now SFedU), defended her PhD thesis on semantic network research at the Institute of Program Systems of the Russian Academy of Sciences (Pereslavl-Zalessky) and her doctoral thesis in the field of graph dynamic models at the Institute of Control Sciences RAS. Author of more than 150 scientific works, including the books Theory of Resource Networks, Resource Networks with Vertex Capacity Constraints, and Chemical Languages of Nervous Systems. Research interests: graph models, activity propagation in networks, heterochemical models of neuronal interaction, models of complex agent interactions.

Dr. Wei Li

Professor, Ph.D., The College of Computer Science and Technology, Harbin Engineering University, China

Multivariate Time Series Prediction in Edge Computing Scenarios and Its Applications

Edge computing, a key enabler of industrial automation and smart cities, enhances real-time responsiveness by processing data near the source. However, its application in multivariate time series prediction faces three critical challenges: resource constraints, dynamic network topologies, and fragmented data streams, which lead to the failure of conventional methods due to rigid static models, poor adaptability, and the trade-off between lightweight design and accuracy. To address these challenges, this study proposes a novel closed-loop framework integrating dynamic spatiotemporal primitives, fragmentation-aware joint modeling, and topology-decoupled inference. The framework provides a lightweight, autonomous solution for industrial IoT, smart grids, and other edge scenarios, balancing efficiency with accuracy while ensuring robustness in dynamic environments.

Speaker BIO

Dr. Wei Li is a Professor at the College of Computer Science and Technology, Harbin Engineering University, where he also serves as Deputy Director of Heilongjiang Provincial Laboratory of New Generation Network Technologies and Information Security. As an IEEE/ACM Member and CCF Senior Member, he holds executive committee positions in CCF's Information Systems/Databases/Human-Computer Interaction Technical Committees and Smart Transportation Special Interest Group. After obtaining his Ph.D. from UNSW's School of Computer Science and Engineering in 2019, he worked as a Research Fellow at the University of New South Wales before returning to China. His research expertise spans database systems, graph data management/mining, spatiotemporal data analysis, and location-based services, with over 60 publications in these areas. He currently serves as Associate Editor for JIHMSP and Guest Editor for Applied Sciences, while actively contributing as PC member for conferences (ICME, ICASSP, DASFAA) and reviewer for top-tier journals (TKDE, TKDD) and conferences (ICDE, VLDB, ACM MM).

Lev Utkin

D.Sc. (Tech.), Professor, Peter the Great St. Petersburg Polytechnic University, Russia

Risk Analysis, Reliability, and Survival Models: Current State, Latest Advances, and Applications in Industry and Medicine

The presentation provides an overview of the current state-of-the-art in reliability and survival analysis models under censored data, where events may go unobserved. The review covers contemporary models based on ensemble machine learning approaches and attention mechanisms. It examines the formulation of survival analysis as a classification problem with imprecise event probabilities. Particular focus is given to challenges in explaining and interpreting model prediction outcomes. The discussion also explores the emerging field of concept-based learning, addressing both concept-based approaches in survival analysis and concept-driven interpretation methods.

Speaker BIO

Professor of the Higher School of Artificial Intelligence in Peter the Great St. Petersburg Polytechnic University, Saint-Petersburg, Russia. Professor, DSc. Head of the Research Laboratory of Neural Network Technologies and Artificial Intelligence in the same university. In 1986 he graduated from St. Petersburg State Electrotechnical University (former Leningrad Electrotechnical Institute). He holds a Ph.D. in Information Processing and Control Systems (1989) from the same university and a D.Sc. in Mathematical Modelling (2001) from St. Petersburg State Institute of Technology, Russia. Awarded an Alexander von Humboldt Foundation Fellowship (2001-2003). Member of the Society for Imprecise Probability Theory and Applications (SIPTA) and the International Society on Multiple Criteria Decision Making (ISMCDM). Author of more than 300 scientific publications, including AI journals: Neurocomputing, Neural Networks, Knowledge-Based Systems, Applied Soft Computing, AI in Medicine, etc. Research interests are focused on machine learning, imprecise probability theory, decision making.

Konstantin Gnidko

D.Sc. (Tech.), Associate Professor, Sirius University of Science and Technology, Russia

Technology for early detection of threats to the information and psychological security of students at educational institutions in the Sirius Federal Territory based on artificial intelligence models and methods

The report presents scientific and practical results of research on the development of technology for early detection and prevention of threats posed by destructive information to students of educational institutions (using the example of the Sirius University of Science and Technology). The technology enables continuous monitoring of multimodal data – physiological signals from wearable devices, speech and facial expressions, activity on social networks and interpersonal interactions in real life, behavioral patterns in ultrametric (p-adic) space, as well as text content – with the aim of early detection of signs of stress or the impact of destructive information on talented young people. The main elements of the developed system are: a stress symptom classifier based on physiological data, deep neural network models for recognizing emotions from voice and facial expressions, social graph analysis algorithms, models for information dissemination across social graphs, algorithms for analyzing text and multimedia content using retrained deep learning models, as well as an ensemble classifier based on the adaptive boosting (AdaBoost) method, which aggregates data from all sources. A mathematical formalization is provided for each component. The work demonstrates that an interdisciplinary approach based on the application of modern artificial intelligence paradigms can significantly improve the level of information and psychological security of students.

Speakes BIO

Doctor of Technical Sciences, Associate Professor, Deputy Head of the Research Group at the Scientific Center for Information Technology and Artificial Intelligence at the Sirius University of Science and Technology. Scientific supervisor of the RSF grant “Technology for early detection of threats to the information and psychological security of students at educational institutions in the Sirius Federal Territory based on artificial intelligence models and methods” and the project “Intelligent technologies for screening social and medical risk factors for sustainable economic development” as part of the third wave of research centers in the field of AI. In 2004, he graduated from the Mozhaisky Military Space Academy. He defended his candidate thesis in cybernetics and systems analysis (2010) and his doctoral thesis in information warfare (2017). Author of over 100 scientific papers, lecturer at the Russian Society “Znanie” , recipient of state and departmental awards. Research interests: machine learning, modeling of cognitive processes, socio-humanitarian aspects of information security.

Pavel Sundeev

D.Sc. (Tech.), Sirius University of Science and Technology, Russia

Cluster model of information security

A cluster model of information protection has been developed, which is an extension of the "classical" information protection model with full disclosure. The cluster model allows us to formally describe protection when solving problems of information systems analysis and synthesis. In the cluster protection model, the problem of accuracy of the formal model is solved by defining the means of protection, subjects and access objects as functional information mod-ules, decomposing information links between modules into physical, syntactic and semantic relationships, taking into account their functional properties essential for information processing and protection, and forming a cluster structure that formally reflects access control policy and constructive protection, followed by an assessment of the security of the static or dynamic topology of the system and the effectiveness of security measures. An approach to the analysis of constructive protection of information systems is shown using the example of distributed registry technology, the possibility of taking into account the impact of access policy architecture and security threats on security. The methods of graph theory, matrix theory, modular cluster networks, and system analysis are used.

Speakers BIO

In 2008, he defended his doctoral thesis in the field of system analysis of the functional stability of critical information systems. Research interests: information security, system analysis.

Vladimir Zaborovskij

D.Sc. (Tech.), Professor, Peter the Great St. Petersburg Polytechnic University, Russia

A turning point in the development of computer science: generative large language models are not all we need

One of the pressing tasks in the development of computer science is increasing the actual performance of hybrid supercomputers operating in shared-use center modes. In the case of choosing an optimization criterion based on a measure of real productivity, it is necessary to find a constructive solution to a number of not only technological but also fundamental problems, some of which do not have a final algorithmic description, for example, fundamental halting problem of Turing machine. The report discusses methods for improving the actual performance of modern hybrid supercomputers based on the use of conceptual machine learning methods and the application of specialized transformers of generative linguistic models, the invention of which became a turning point in the development of computer science. The report shows that the usage of linguistic models without consideration of conceptual requirements and explanation of the generated results can lead to erroneous conclusions.

The main idea behind the proposed approaches is to use the results of solving a flow of applied problems marked by the supercomputer dispatcher. The utilization of conceptual models for managing supercomputer resources enables an increase in the number of successfully completed application tasks, while the application of attention, self-attention, and multidimensional attention mechanisms improves the accuracy of task completion time estimates by taking into account contextual conditions represented in the form of personal models of user competencies. The report shows that the actual performance of supercomputer platforms can be improved by using exo-intelligent memory-oriented computing architecture with a built-in machine learning mechanism and storage of data on the results of calculations performed. This architecture can be considered as an alternative to the traditional approach to extensive growth of computer resources based on the use of specialized graphics accelerators and programmable logic integrated circuits.

Speakers BIO

Professor at the Higher School of Artificial Intelligence, Peter the Great St. Petersburg Polytechnic University, St. Petersburg, Russia. Professor, Doctor of Technical Sciences, Head of the Research Laboratory for Supercomputer Technologies and Machine Learning. In 1979, he graduated from Saint Petersburg Polytechnic University (formerly Leningrad Polytechnic Institute). He defended his candidate thesis on solving adaptive control problems based on the identification of models of large power systems (1984) and his doctoral thesis in the field of information transfer process control and fractal process control in TCP/IP networks (1999). In 2010, he became a laureate of the St. Petersburg Government Award in the field of science and education. From 2008 to 2015, he was the scientific director of a series of the "Kontur" space experiments for telematic control of robots from aboard the International Space Station, and from 2021 to the present, he has led research in the field of machine learning for resource planning systems for hybrid supercomputers operating in shared-use mode. Research interests: machine learning, computer science theory, decision-making methods.