No 4 (2023)

We consider a class of multicriteria choice problems in which the preferences of the decision maker are modeled by an interval type-2 fuzzy relation. The basic axioms of ‘reasonable’ choice are formulated. They, in particular, allow us to establish the Edgeworth-Pareto principle for this class of problems. The concept of a quantum of interval fuzzy information is introduced, as well as a consistent set of similar quanta. A criterion for the consistency of a set of quanta is formulated and a scheme for using quanta of interval fuzzy information to reduce the Pareto set is presented. An example is given to illustrate the proposed approach.

The paper describes the concept and information model for recognizing, analyzing and assessing the interaction of people from different target groups based on emotional reactions. The model includes classification and identification of people by the cloth color, tracking their movement in a limited space, determining the fact of interaction between people and assessing the quality of interaction by human emotions, which are determined by face images and voice sounds. The concept and information model are realized in the developed automated video surveillance system for human interactions. The results of testing the automated system are presented.

This article is devoted to the use of modern language models based on BERT and models based on three types of text linguistic features for automatic determination of the text genre, as well as a comparative analysis of these models from the points of view of computer and classical linguistics. The authors have collected their own corpus of Russian-language Internet texts in eight genres: VKontakte posts, comments, articles from the Habr portal, retail descriptions, news, scientific articles, advertising, movie reviews from the Kinopoisk website. Each text was represented as a vector of numerical features using each of the selected models: five BERT variations and linguistic features of character, structure and rhythm levels. Vectors based on linguistic features were also concatenated for two or three levels to obtain additional text models. Next, the vectors were classified into eight genres using neural network classifiers, a perceptron and LSTM. The results of the classification showed that BERT models achieved a high quality of genre detection: up to 91-99% of precision, recall, and F-measure. The combination of linguistic features made it possible to obtain the F-measure about 90%. An analysis of the classification results and text models from a linguistic point of view revealed the features of individual genres and possible reasons for both high results and classification errors.

The paper presents a scientific biography of the famous scientist and organizer of science, G.S. Osipov. The range of his research interests is outlined and the main results of fundamental and applied research into models and methods of artificial intelligence are presented. The contribution of G.S. Osipov and his scientific school in various areas of development in the field of artificial intelligence is characterized.

The paper proposes an integrated approach to the selection of technological solutions for ore preparation, which allows combining intellectual and quantitative methods to justify decisions on the management of mining and mineral processing. The intelligent system includes a database for making optimal decisions, models based on neural networks and classical mathematical methods, technical means, technological operations and organizational techniques that allow for ore quality management measures. A new approach to the classification of fragments of ore fractions is proposed, based on a neural network that has the functionality of finding ore mineral grains in an image with a subsequent assessment of the degree of its disclosure, which made it possible to increase the efficiency of image recognition of ore sections by at least 5% compared to the analytical method of ore analysis.

In the first part of the article, the necessity of transition from subject to spatial methodology is substantiated, a convergent approach is proposed for convergence of the properties of the "intelligence of space" into its rules. In this article, with the help of a convergent approach, the convergence of the manifested properties of the "intelligence of space" (gravitation, mutual influences of the geometry of space and matter, emergence, etc.) into the rules of spatial methodology is carried out. The existing essences of geometric forms of the rules of elementary distinction and the rules of their selfdevelopment are considered. The initial essence of the geometric form of the rule of artificial preparation of the emergence effect and the process of its materialization into the original semantic essence of the graphical algorithm for the implementation of an elementary action are proposed. The process of self-development of the initial essence of the graphical algorithm of elementary action into a meaningful graphical algorithm for overcoming the gap or solving problems is shown. In the process of implementing the algorithm, the meaning will be transferred to the target matter.

The paper demonstrates computational efficiency of probabilistic approach to knowledge extraction through binary similarity operation. In addition to previously proved by the author the result on sufficiency of a polynomial number of hypotheses on causes of investigated target property, the paper contains a polynomial upper bound on mean working time of the algorithm to generate a single candidate for hypothesis. The proven result concerns a family of algorithms based on coupled Markov chains. To obtain a good estimate for the length of the trajectory (before entering the ergodic state) of such a chain, we needed to enrich the training sample by adding negative columns for existing binary features.

A modified nonparametric algorithm for automatic classification of large-volume statistical data is proposed. Its application makes it possible to detect classes corresponding to unimodal fragments of the probability density of a multidimensional random variable. The compression of the initial information is carried out on the basis of the decomposition of the multidimensional space of features into a data array composed of the centers of the sampling intervals and the corresponding frequencies of belonging to the values of the random variable. Based on these data, a regression estimate of the probability density is synthesized. The information obtained is the basis for the algorithmization of the automatic classification procedure. A class is a compact group of observations of a random variable corresponding to a single-modal fragment of probability density. The computational efficiency of the modified nonparametric algorithm for automatic classification of large-volume statistical data is provided by the compression procedure of the source data, improvement and algorithmization of the traditional nonparametric method of class detection. The computational efficiency of the modified non-parametric algorithm for automatic classification of large volume statistical data is provided by the initial data compression procedure, improvement and algorithmization of the traditional nonparametric method for detecting compact groups of observations of a random variable. The effectiveness of the developed method of automatic classification is confirmed by the results of its application in the analysis of remote sensing data of forests damaged by the Siberian silkworm.