News of the Kabardino-Balkarian Scientific Center of the Russian Academy of Sciences

The article presents development of a machine learning model for predicting fraudulent transactions using transactional data from a bank. It discusses the features of encoding categorical variables related to the presence of time in the transactional data to avoid information leakage. Additionally, experiments were conducted on the application of bagging and the creation of additional variables based on their contribution to the final prediction using Shapley values. The quality metrics of the machine learning model are examined and analyzed.

The digital transformation of a business means the digitization of many processes in an enterprise, i.e. it assumes the implementation of processes using computer technology and IT technologies. At the same time, it is important to organize the effective integration of existing processes in the enterprise with modern IT technologies. Such integration may concern not only production, but also other areas of human activity. Of course, many industries have been automated to varying degrees before, but the advent of artificial intelligence (AI) can smooth out the difference between industries with varying degrees of automation and optimize processes, even if some of the fields of activity do not involve the use of AI. Nevertheless, the process of digitalization in the vast majority of cases will accelerate decision-making if AI systems are used, in particular a digital twin. This optimizes data collection, which will allow them to be used to create models of objects or systems. In the future, the model will be used to analyze and optimize work without the physical presence of an object. All of the above determines the relevance of the topic of determining the role of artificial intelligence in the transformation of Russian business. In this article, the authors reflect on the problem "What is needed for the development of new data analysis technologies in production? And how can we improve the data environment?" The article provides an overview of the history of the use of artificial intelligence in business. The weaknesses of using artificial intelligence technologies are discussed. An attempt is being made to answer the question of what needs to be done today so that an enterprise or organization can take a leading position tomorrow.

Сomputer modeling of weather conditions based on atmospheric pressure data is carried out шn the article with the help of the theory of fuzzy sets. An algorithm was implemented for calculating integral ranking indices for fuzzy sets that characterize weather conditions in the computer mathematics program Maple2021, with the help of the LinearAlgebra library. It was shown that if the atmospheric pressure is not very high, the next day will be "sunnier", than in the case when the pressure is "very low". It was confirmed that integral ranking indices give more accurate information than deterministic ranking indices.

This paper reviews state-of-the-art reinforcement learning methods, with a focus on their application in dynamic and complex environments. The study begins by analysing the main approaches to reinforcement learning such as dynamic programming, Monte Carlo methods, time-difference methods and policy gradients. Special attention is given to the Generalised Adversarial Imitation Learning (GAIL) methodology and its impact on the optimisation of agents' strategies. A study of model-free learning is presented and criteria for selecting agents capable of operating in continuous action and state spaces are highlighted. The experimental part is devoted to analysing the learning of agents using different types of sensors, including visual sensors, and demonstrates their ability to adapt to the environment despite resolution constraints. A comparison of results based on cumulative reward and episode length is presented, revealing improved agent performance in the later stages of training. The study confirms that the use of simulated learning significantly improves agent performance by reducing time costs and improving decision-making strategies. The present work holds promise for further exploration of mechanisms for improving sensor resolution and fine-tuning hyperparameters.

The article presents the results of studies on the rooting of tall blueberry cuttings conducted in 2022–2024 in the foothill zone of the KBR (LLC "Yug Agro"). Layering, when in contact with a wet peat substrate, is capable of rooting directly in the field. The advantage of this method of reproduction is its simplicity, which does not require the use of expensive technological elements, a high percentage of rooting, as well as the absence of the need for additional cultivation for planting in a permanent place. The purpose of the study – is to optimize the method of propagation of tall blueberries by layering in the conditions of the foothill zone of Kabardino-Balkaria. Research methods. The furrowing of the bark of the part of the shoot buried in the ground was carried out with a knife with teeth. In variants with stimulation of root formation with Phytactive Extra Plus, the gel was applied to the buried part of the shoot after furrowing (or without it) with a brush. Each experimental version was laid and accounted for in three repetitions of 100 accounting layers in each repetition. Research results. It was found that furrowing the bark and subsequent treatment of this area with Phytactive Extra Plus gel before the shoots are sunk into the ground ensures the rooting of 81% of the layers. At the same time, the roots are strong, white, well-branched, and penetrate into the soil by an average of 16 cm. This ensures that the planting material is suitable for planting in a permanent place without additional regrowth during one growing season.

The article presents the results of research on one of the most pressing problems – the study of the dynamics of phosphorus reserves in the soil under conditions of intensive use of arable land, including the main factors influencing the change in phosphorus content, including agrotechnical practices, fertilization, soil erosion and biological activity. The purpose of the study is to determine the dynamics of phosphorus reserves in the soil due to intensive use of arable land. The scientific novelty lies in the fact that special attention is paid to the effects of long-term agriculture on phosphorus balance and soil fertility. The experiments, observations and records were carried out according to the methodology adopted in agronomy. As a result of research, it is important to note that the phosphorus content for this period under alfalfa corresponds to the content of P₂O₅ under annual crop crops. In a 10–40 cm soil layer, the phosphorus content turns out to be almost equal under all crops and black steam, but 23–12% less than in a field undergoing semi-steam treatment. In the 40–100 cm soil layer, the P₂O₅ content is leveled under alfalfa, black fallow, and chilly and is 14–28% less than in crop crops of annual crops, i.e. the opposite pattern is observed compared to the overlying soil layers, which is due to the redistributing role of the root system and the decomposition of crop and root residues. And according to the dynamics of P₂O, depending on the method of using arable land and its precursors under the rye mixture, it can be said that the powerfully developed alfalfa root system increases the mobility of phosphates. During the analyzed period, only the alfalfa root system in the 40–100 cm soil layer supports phosphorus activity. In fields unoccupied by plants, phosphorus binds to sedentary forms. A similar pattern is typical for the entire meter-high soil layer. According to the results of the conducted research, it can be said that phosphorus, which is part of the crop and root residues of crops previously growing in the field, is insured against possible transformations in the soil and, like contained in manure, is easily accessible by this time to a new generation of plants (winter – basic and intermediate, corn and Therefore, the retention of N, P₂O₅, K₂O and other macro- and microelements from fertilizers and soil in the crop-root sediments of plants is a well-known phenomenon, but not yet actively used.

Sustainability in digital logistics networks is a complex issue that encompasses economic efficiency, social responsibility, and environmental preservation. Due to this, there is increasing demand to address logistics operations with the concepts of digitalization and sustainability. Despite the rapid development of the logistics industry, there is still a research gap concerning the transformation of logistics to sustainable digital logistics. Therefore, this article aims at the review of empirical and methodological frameworks of the current state of studies on the digital logistics network and sustainability with particular emphasis on, but not limited to, agricultural logistics. Reputable academic sources were selected for the review and comprehensive analysis of the wide range of tasks. The review result identified some innovative methods, robust models, and explored certain significant factors. Merits and demerits of the review sources were assessed. Future research in this area could incorporate empirical analysis with a large scope to analyze the logistics operations of different sectors perspectives on digitalization and sustainability.

The relevance of the topic is determined by the importance of addressing logistical issues in the context of the global growth of the biofuel industry, the increased need for sustainable management of logistics processes, and the reduction of the carbon footprint. The development of integrated logistics solutions is particularly timely, as it enables the consideration of rapidly changing market demands and environmental standards. Research Gap. Currently, existing approaches to optimizing multimodal logistics have significant shortcomings related to the unsynchronized management of information and material flows. In addition, there is a lack of empirical data on the integration of omnichannel methods, among which the following are applied: Digital planning using artificial intelligence algorithms; Carbon emission monitoring; Optimization of intermodal (multimodal) transportation. Research Objective. The objective of the research is to develop an optimization model for an omnichannel logistics network for biofuel, based on data analysis methods and artificial intelligence. This approach enables the creation of an effective tool for managing Russian-Chinese logistics networks in a cross-border context. Scientific Novelty. The data-driven optimization model developed significantly reduces logistics costs, cuts carbon emissions, and enhances the resilience of the supply chain. This approach expands the theoretical foundations in the field of omnichannel logistics and opens up new prospects for the use of modern digital technologies in optimizing transportation systems. Scientific Discussion and Future Research Directions. The authors propose to discuss the possibilities of adapting the suggested model to solve similar logistical challenges in other sectors of the economy. An important direction of the discussion is also the improvement of organizational and economic mechanisms for the integration of digital technologies into the logistics system, particularly the refinement of carbon emissions monitoring methods, which will enhance the overall efficiency of optimizing logistical processes.

This article highlights the mechanisms through which corruption affects key economic indicators and the business environment. The author examines how corruption risks transform the investment climate, create additional costs for business and impede the implementation of long-term projects. The study identifies the specifics of the impact of corruption schemes on small and medium-sized enterprises and traces the link between corrupt practices and the emergence of shadow structures in the economy. In addition, the paper analyses the impact of systemic corruption on the allocation of budget funds and the evaluation of strategic state programmes. Based on a comparative analysis of scientific sources, the paper offers specific recommendations for the implementation of anti-corruption measures that increase the transparency of market procedures and stimulate the inflow of investments. The conclusions formed may be in demand by economic specialists, representatives of public administration and experts involved in the study of socio-economic reforms.