Software & Systems

Articles of journal № 2 at 2022 year.

Order result by: Public date | Title | Authors

11. Evaluation of the effectiveness of chemical reaction conditions [№2 за 2022 год]
Authors: N.V. Zvyagintsev, Billig V.A.
Visitors: 1824
The paper considers the problem of evaluating the effectiveness of chemical reaction conditions taking into account such factors as the presence of impurities, the cost of catalysts, and some other factors af-fecting the cost of a technological process. In order to evaluate the effectiveness of a chemical reaction, this paper proposes first to inde-pendently evaluate the effectiveness of each factor involved in a reaction, and then build a summary estimate that takes into account the effectiveness of each factor. Since the nature of factors is different, the authors introduce the concept of bonuses awarded to each factor in order to be able to compare the influence of factors. Bonuses are awarded for receiving the main product, as well as for minimizing a by-product. Using the example of such factors as pressure and temperature that affect the reaction condition, the paper shows the expediency of introducing a concept of “soft condition”. A soft condition value is a value at which the costs of its implementation are minimal. Taking into account these assumptions, the evaluation of the effectiveness of each factor is constructed as a fuzzy measure of efficiency – a mono-tone function with values from the interval [0,1]. One of the approaches for assessing the significance of a particular factor is based on the possibility of using data mining methods. This method assumes the possibility of accumulating a sufficiently rep-resentative database. The total efficiency score is constructed as a weighted sum of the estimates of each factor. The accuracy of the proposed approach was verified on the real experiment data while recording both factors affecting the course of the chemical reaction and the amount of target and by-product ob-tained as a result of the reaction.

12. Software implementation of demographic data analysis based on the unified population register [№2 за 2022 год]
Authors: Yusifov F.F., Akhundova N.E.
Visitors: 1728
A unified population register is a key component of the e-demographic system. The register is based on the integrated databases exchanging both aggregated data and individual data between separate regis-ters. The paper examines the analysis of demographic data on the basis of a unified population register. Population registers play an important role in obtaining information about the population. It should be noted that the COVID-19 pandemic has once again emphasized the importance of using administrative data as e-registers for demographic research. The paper provides an experimental analy-sis of demographic characteristics in the context of the COVID-19 pandemic based on the data of indi-viduals integrated into a unified register. The data on individuals in the study are hypothetical data tak-en from two separate registers: the population and health registers. A database was taken for 1000 peo-ple integrated into the unified register. The paper presents the program implementation of demographic data analysis. Demographic analy-sis was implemented in Jupyter Notebook 6.1.4., Python 3.8.5. The results show that the establishment of an e-demographic system requires the integration of various state registers for more detailed analy-sis. This will allow processing and analyzing larger and more multidimensional structured data at dif-ferent time intervals. At the same time, the reliability of the information included in the register, the elimination of inconsistencies, and ensuring continuous updating of registration information for each individual are very important issues. Elimination of errors in registration data makes unified popula-tion registers a reliable source of information.

13. Developing universal framework design for federated learning [№2 за 2022 год]
Authors: Efremov M.A., I.I. Kholod
Visitors: 2403
The paper researches the technology of federated learning that allows collective machine learning on distributed training datasets without transferring them to a single central storage. The relevance of the technology is determined by the long growing trend towards using machine learning methods to solve many applied problems on the one hand, and by the growth of requests for privacy and data processing closer to the data source or directly at the source, including legislative ones, on the other hand. The main problems in creating federated learning systems are the lack of flexible frameworks for various federated learning scenarios: the majority of the existing solutions focus on training artificial neural networks in a centralized computing environment. The subject of the research is the common framework architecture for developing applied federated learning systems, which allows building systems for different scenarios, parameters and topologies of the computing environment, various models, and machine learning algorithms. The article considers the federated learning subject area, gives the main definitions, describes the process of federated learning, presents and analyzes various scenarios of possible applied tasks for federated learning. It contains the analysis of the most well-known federated learning frameworks at the time of writing, as well as their application for possible cases that were described previously. As a result, there is a description of the architecture of a universal framework that, unlike the existing ones, can be used to develop applied federated learning systems of various types and different cases.

14. Development of decision support programs based on Bayesian probabilistic models [№2 за 2022 год]
Authors: G.I. Kozhomberdieva, D.P. Burakov , Khamchichev G.A.
Visitors: 3550
The paper presents programs focused on using decision-making support tools and implementing origi-nal approaches to a rating estimation by an expert group and to fuzzy inference. The programs use the probabilistic models based on Bayes' formula previously proposed and pub-lished in the works of the authors. The models interpret the input estimated data as evidence in favor to one or another hypothesis from a set of possible ones determined by the model specifics: hypotheses on the object’s place in the rating (in the group expert rating estimation model) and hypotheses on the possible value of the output linguistic variable (in the fuzzy inference model). The obtained evidence is transformed in a model-specific way into a set of Bayesian conditional probabilities computed under the assumption that the corresponding hypothesis is true, and then poste-rior probability distributions on the set of these hypotheses are calculated. These posterior distribu-tions are used to obtain the final result: a rating of objects (in the rating estimation model), defuzzified value of linguistic variable (in the fuzzy inference model). The paper discusses the features of the software implementation of models on Java platform, notes the advantages of models confirmed or identified in the process of software implementation. The de-veloped programs are registered in the Register of Computer Programs of the Russian Federal Service for Intellectual Property (Rospatent) and are used in the educational process at St. Petersburg State Transport University.

15. Development of a prototype solver for extended step theories of propositional logic [№2 за 2022 год]
Authors: Fominykh I.B., Alekseev N.P., N.A. Gulyakina, Kravchenko K.S., Fomina M.V.
Visitors: 2820
Nowadays, there are active researches on the possibilities of using non-classical logics in modeling the cognitive agent’s reasoning. The paper considers the problem of developing and implementing a prototype of an Extended Step Theory solver (EST) in the case when decisions on managing a complex technical object are made un-der strict time constraints. The authors consider a logical system based on using step theories with two types of negation, such systems are called EST. The use of two types of negation allows deducing both unbiased facts and belief facts, which is important when modeling human reasoning. The paper focuses on the issue of organizing the inference procedure based on using non-classical logics in modeling the reasoning of a cognitive agent. There are the main stages of the development of the EST prototype using the propositional logic lit-erals given. There are also descriptions for each solver component, its functions, tasks, input and out-put data. The authors is justify the choice of the clingo output system supporting the formation of ex-tended logic programs Answer Set Programming (ASP) as a tool for implementing the solver. The paper gives the algorithms of translating the EST into a logical program corresponding to the ASP syntax. When organizing logical inference, the authors used the algorithm of cyclic processing of EST belief sets in the clingo environment. The main stages of this algorithm are considered by an example that analyzes the solver’s operation stages and the presents the results in the clingo syntax. An example of the solver's work demonstrates the main EST features in hard real-time problems, such as the rejection of logical omniscience, self-knowledge and temporal sensitivity. It is planned further to consider the applicability of the created solver to a more complex formal system – the logic of first-order predicates.

◄ ← Preview | 1 | 2