|
The Bulletin of the Adyghe State University, the series “Natural-Mathematical and Technical Sciences” |
 Đóññêèé
|
| Home page Subjects Chief Editor Editors Instructions to Authors Rules of the reviewing Subscription Contacts Publishing house | ||
|
#4 / 2022Biological SciencesBibalova L.V., Zaytseva S.A.Species composition of riverside vegetation of the smooth terrain in the Adyghea Republic
The paper discusses the results of a study on the species composition of riverside vegetation of the flat land of the region. We identified 104 species belonging to 23 families. The average num-ber of species on an area of 49 m2 varied from 13 to 20 and averaged 18 species. This indicates that a floral diversity of phytocenoses of the Ulka River Valley is low. During the study, we identified five main associations and indicated all the species that formed plant communities of river-side vegetation.
The impact of vehicle emissions on the environment of the city of Maikop, Adyghea Republic
The paper discusses the results of a study on the impact of vehicle emissions on the roads of the city of Maikop. The streets most polluted by motor transport are Proletarskaya – 1242 cars, Deputatskaya, Pionerskaya – 1208 cars, and Pervomayskaya – 980 cars, atmospheric air pollution with carbon monoxide reaches 42.0%.
MathematicsShumafov M.M., Panesh T.A., Havadja M.A.On the stochastic stability in the large of solutions of the nonlinear second-order differential equations perturbed by white noise. V
This paper is a continuation of the previous paper and presents the fifth part of the authors’ work. In the first four parts of the work, we presented some preliminaries from probability theo-ry, stochastic processes, theory of stochastic differential equations and stability theory of stochastic differential equations. In the work, nonlinear autonomous second-order differential equations under random perturbations are considered. Here, in the paper we give new sufficient conditions for stochastic asymptotic stability of solutions of the nonlinear second-order stochastic differential equations. The probabilistic estimations of the remaining of a random trajectory in a bounded region of the phase plane are given. As an example, harmonic oscillator perturbed by white noise random process is considered.
Technical SciencesRybanov A.A.CRUD model of security as a basic element of data access control management in information systems
The CRUD model is one of the ways to describe the relationships between data processing and data elements with respect to their creation, reading, updating, and deletion. The article pro-poses expanding the two-dimensional representation of the CRUD model to a three-dimensional CRUD security model by introducing an additional dimension describing the rights and privileges of database users. The implementation of the CRUD model of security based on ROLAP technology is considered. The approach proposed in the article is focused on access control management processes within the framework of database development and protection technologies.
Analytical review of methods for assessing the quality of classification algorithms
The article discusses the main methods for evaluating and analyzing the effectiveness of various classification algorithms with a description of the features of these methods and examples of their use. The publication provides the definition of the confusion matrix and an extensive set of different metrics basing on the data of the error matrix. The work describes the metrics for the case of binary classification, and then gives an extension of the estimation technique to the case of several classes.
Analysis of the implementation of federated learning in boundary computing using the FL protocol
The purpose of this study is to review and analyze the application of a relatively new machine learning technique in a distributed information collection and processing system – artificially intelligent boundary computing, which has proven itself well for the implementation of a swarm of drones performing independent flight in terms of transmitting collected data and training an artificial intelligence model. Special attention is paid to one of the paradigms of this technique – federated machine learning.
Using diffusion models to develop applications that generate images based on text queries
The purpose of this article is to consider diffusion models, as well as to explore the possibilities of using neural network data to develop applications that generate images based on text que-ries, and to compare the diffusion models that exist today. Diffusion models are a subcategory of deep generative models, which consist of stages of forward and reverse diffusion, generate data similar to those they use to train. Different architectures of diffusion neural systems can be used both to generate images based on text queries and to transform existing images. In the perspective of the development of diffusion neural networks, we assume that their use will significantly facilitate the work of designers in production.
Design of an actual task planning solution based on the Android operating system
This article discusses the design of a modern mobile application for planning. The relevance of the work lies in the high practical value of mobile applications for planning. By developing their more advanced counterparts, users have more opportunities to carry out planning, and there-fore improve the quality of their lives. The article also reviews existing application functionality and suggests new ones as an improvement. The publication substantiates the choice of certain tools for implementation and provides the design of the application architecture, both the con-nection between the design patterns used and the implemented third-party components. The au-thors describe the design of the program database and the calendar window, both from the point of view of the implementation of the previously laid architecture, and from the interface. Additionally, the article presents a model of the notification generation algorithm, taking into account the selected deadlines for completing the task, its importance and the overall workload of the user. The main research methods are analysis and modeling. We can use the results of the study in the design and implementation of a mobile application for planning.
The approach to solving the problem of quality control in the service sector based on the construction of a data mining system
This article discusses a new approach to quality control in the service sector by implementing software based on intelligent systems, including, first of all, methods of processing and analyzing natural language using data regularization. The article also presents the results of a comparative analysis of several, the most commonly used, neural networks-transformers: BERT, GPT and XLNet. In the course of the study, we have revealed that the BERT algorithm, thanks to bidirectional learning, is more suitable for processing input information; the GPT model, due to autoregression, for generating system responses; the XLNet method, thanks to the PLM mechanism, for language modeling. The research shows that each of the considered algorithms has its advantages and disadvantages, therefore, in order to achieve optimal results, we should use software complexes based on several transformer algorithms.
|
© Adyghea State University, 2005-2023