其他名称：
25-21-00029
项目负责人：
Tselykh Alexander
发表日期：
2025
主持机构：
Southern Federal University,
国家：
俄罗斯
开始日期：
2025
结束日期：
2026
简介：
The project is aimed at solving the fundamental problem of the development of convergent technologies - modeling the mechanisms of mental activity of a decision maker at the strategic level and developing an artificial intelligence (AI) system that replicates the process of making strategic decisions by a person based on abstractions when managing complex socio-economic systems.
The urgency of solving the indicated problem is determined by the ever-increasing complexity of the tasks being solved in almost all areas of modern science and the real economy. This predetermines the indicated direction for the development of AI systems. At present, there are only the very first attempts to solve this problem, which is important for decision-making in political science, economics, military science, etc., in the problems of predicting the strategic reaction of decision makers (DM) during planning and counterplanning.
Our approach is that the basis of cognitive processes of the highest level is the general idea of a person about the system in relation to which a certain decision is made. Following in this direction, we consider this problem in the context of the phases of the theory of thinking (strategy → action) as part of a strong artificial intelligence system, when people first solve the problem of choosing goals, and then the task of implementing them.
The scientific importance of the problem identified in the project is determined by the fact that the theoretical principles that represent the process of replicating the decision-making process of strategic decision makers in the management of complex socio-economic systems with unobservable variables based on causal-cognitive models will be the basis for the robotization of this process as part of the cognitive intelligence system.
The problem of modeling the mechanisms of mental activity of a DM at a strategic level in an AI system that replicates the process of making strategic decisions by a person based on abstractions when managing complex socio-economic systems with unobservable variables of a non-stochastic nature has not been solved before and is new th.
The approach proposed in the project is based on the implementation of this paradigm through the concept of the causal-cognitive model (CM). This concept is based on the assumption that our brain performs a functional transformation of the knowledge acquired by a person during the period of into a cognitive map, that is, the brain "compresses" knowledge, building its logical structure and forming a mental picture of the world in the DM. This mental picture is represented by us in the form of a graph model, in fact, representing an encoded concentrated knowledge of the decision maker. Based on this mental picture, the decision maker makes decisions at the strategic level.
The scientific novelty of the proposed project is to achieve the following scientific results:
1. A new approach to the implementation of the AI system, in which the functions of perception are expressed through entities and their relationships (in the form of CM), and external semantic information is entered through access to a comprehensive world knowledge base (Web search).
2. An original cognitive architecture based on a hybrid approach, the center of which is CM.
3. A new approach to learning a non-stochastic CM with unobserved variables based on the representation of the learning environment as a spectral decomposition of the generated adjacency matrices with a given direction corresponding to the nature of the system being modeled.
4. Expansion of the concept of the state of a system with unobservable variables by integrating the variables and parameters of the control model, and representing the unknown variables of the model (impacts and responses) in the form of their spectral mappings.
5. A new approach to the implementation of the process of making strategic decisions, based on the built-in global optimization, which implements the law of correspondence of the decision to the cumulative advantages and disadvantages of their choice in CM with unmeasured parameters.
专业领域：
信息技术
语种：
英语
