The concept of an intelligent educational platform of a new type is proposed. This platform helps students independently assimilate disciplines within the interaction with an electronic learning system that adapts to the students’ basic level of knowledge, psychological type, preferences and individual characteristics of the nervous system. The basis of the concept consists of integrating an electronic textbook on educational disciplines, based on the principles of programmed education, and gamification, implemented in the web development of the shell of an intelligent educational platform.

The resistive grounding of the neutral is used in power distribution networks and has several advantages over the insulated one. As a rule, transition resistance is occurring at a fault location during a single phase-to-ground fault. The value of the resistance can vary widely. The transition resistance effect on the processes occurring during a single phase-to-ground fault is studied. The article presents the results of simulating the mode in the network with resistive grounding of the neutral at various transition resistance values.

The article presents the practice of effective application of CDIO principles and functional possibilities of the eduScrum method when managing students’ online team project activity in the field of IT.

The article discusses the issues of data pre-processing and visualization for internet marketing using such technologies as Python (front-end processing of data received from suppliers and formation of datasets for reports) and PowerBI (visualization of data for further interpretation and decision-making purposes). The problem of analysis of efficiency of website pages promotion in Google search was solved. A set of key performance indicators (impressions, clicks, CTR, an average weighted position, the amount and rate of conversions, the amount of profit etc.) and the fuzzy logic paradigm (detection of pages preferred for promotion according to the pre-determined criteria) were used as methods for the implemented algorithms.

The study describes an implementation of an algorithm for a systematic research using methods of mapping, visualization and network analysis of complex networks arranged from a set of forecasting terms, which were extracted from bibliographic databases. The proprietary method for selecting
terms based on a matrix that determines the level of maturity of self organizing intelligent systems is used. The visual clustering analysis of the map of terms related to the advanced information technologies by the subject matter and downloaded from Scopus databases for the period of 5 years was carried out. The analysis of a general state of the network and that of the network connections by individual clusters is carried out. The study investigates properties of the networks aimed at detecting percolation processes in breaking nodes connections both in decreasing order of their centrality and in random order. The key role of central vertices in the processes of information dissemination in the terminological network is substantiated by the obtained empirical data. The study shows that the connections evidently disintegrate if there is a combination of nodes with low centrality indicators (< 15), i.e. the percolation threshold is overcome when approaching the indicator of ≥ 15 % of network nodes activation.

The article determines and formulates the main requirements to a neural network that detects hand gestures. A study of existing approaches is conducted on the matter of their compliance with the requirements. A list of real conditions is compiled to imitate an application of a sign recognition module “Dialogue Sign Language Translator”. In addition to that, the required camera characteristics were measured for proper operation of the neural network in order to obtain a real-time palm recognition quality of at least 75 %. A prototype of a neural network was developed. The study shows the efficiency at differentiating left and right hands during imitation. From the prospect point of view, it is necessary to build the software operating both with signs and with speech in order to continue developing the technologies for inclusion of people with communication difficulties.

The article proposes a method for simulating and implementing a team activity when designing an automated control system for the “Local farm” complex. E-network logical and dynamic model was used to group project teams and to assess in-time and high-quality work performance. The success of the interdisciplinary approach, methods of decomposition of the project into steps and delegation of the tasks to the teams’ participants is shown. The efficiency of adaptation of project decisions when creating a pilot version of an automated control system for the “Local farm” complex is presented.

The article discusses the problem of classification of news texts in Russian using such machine learning algorithms as naive Bayes classifier, random decision forests, logistic regression, and artificial neural network. The texts of the Internet news portal Lenta.ru were selected from nine different classes –
sections of news articles to solve the problem. The software implementation in the framework of the study was carried out using the Python programming language. The preprocessing of text data included removal of irrelevant characters and their reduction to a common register, tokenization, normalization, removal of stop words and vectorization of texts. Tensorflow and Keras libraries of the Python programming language were used to implement an artificial neural network. For each of the machine learning models used, hyperparameters values were determined in order to achieve the highest classification quality using a number of metrics: precision, recall and F-measure. A comparative analysis of the algorithms used was carried out. Possible ways for further study within the problem in question are specified.

The article proposes a method for the automatic classification of time series responding to normally and pathologically functioning arteriovenous fistula in patients during hemodialysis. The method is based on the hypothesis that a normally functioning fistula blood flow is laminar, and a pathologically functioning one is turbulent, as well as on its analogy to the mathematical problem on differentiation of regular and chaotic time series. Two methods were applied to solve the problem. The first method involved finding the series position in the entropy-complexity plane. Following that, the specified position was compared with the identified clusters of values of a time series set. Proposed by the authors, the second method involved constructing an object-attribute graph within the framework of the theory of formal concepts analysis. Both methods proved to be effective in determining the fistula state, with the second method being more efficient for detecting thrombosed fistulas.