2023 year, volume 27, issue 2 (PDF)

Ryjov A.P., Gorny B.E., Zudin A.V. Measurement of alcohol well-being in regions based on statistical information

To reduce the risks associated with alcohol abuse, it is extremely important to assess alcohol consumption levels. At the same time, it is not enough to know indicators that assess only the volume of retail sales, which does not take into account unrecorded alcohol consumption. The development of an integral indicator of the index, based on available statistical data, will make it possible to level the risks of expert assessment and increase the efficiency of spending budget funds in the field of healthcare. The report provides a description of the available expert and factual information, the structure of the index, an approach to its assessment by means of the theory of fuzzy sets is proposed. The features of the implementation of the index calculation in the Matlab environment are considered, examples are given. The related direct and inverse problems are formulated and discussed, the solution of which will optimize the efficiency of the healthcare system in this parameter at the regional and federal levels. The alcohol well-being index is one of the important parameters of the health care system, on the one hand, and a typical index of socio-economic processes, on the other, therefore, the approaches described in the report can be applied to construct a wide set of such indices.

Keywords: preventive medicine, alcohol well-being, process evaluation and monitoring.

Khusaenov A.A. Learning parameters rotation method

A method is proposed for improving the quality of training of convolutional artificial neural networks (ANN) by dividing the parameters according to their ability to expand the receptive field. For example, ResNet50 accuracy increase is achieved by only 4 layers freeze. It is shown that the model generalizing ability increase is achieved by eliminating the excessive contribution of individual significant (occlusive) image elements in the feature maps. In favor of these assumptions the results of experiments in the transfer learning task and reasoning about the existence of this problem are presented. The proposed approaches can be useful in training ANNs on small data or distillation of the training set, where the problems of overfitting on individual occlusive features are of high importance.

Keywords: convolutional artificial neural network, neuron receptive field, model overfitting problems, feature occlusion in convolutional artificial neural networks

Alekseev D.V. To the question of restoring an image from an erased code

The article proposes an algorithm for image recovery by erased code, i.e. the code without information about indices. The algorithm has a polynomial complexity. The definition of general position triangles is given. The algoritm correctness is proved for wide class of points set which form general positions triangles.

Keywords: image code, image encoding, erased code, affine equivalence.

On the complexity of converting to a correct linear form Dergach P.S., Saltsova D.A.

This paper is devoted to the study of the regular linear form for regular languages with polynomial growth function and improving the corresponding estimates on the complexity of the transition from the linear form to the regular linear form. We managed to lower the previously known estimate \[ n^2 \] from [1] to an estimate \[ \frac{n^2}{2} + n \]. Also obtained an upper estimate \[ \frac{n^2}{4} \] for languages of the form \[ \beta^*_1 \beta^*_2 ... \beta^*_s \], in which the words \[ \beta^*_i \] are commensurable.

Keywords: regular language, growth function, regular linear form, complexity.

Nosov M.V. Estimates of the degrees of separating polynomials for monotone and self-dual functions

In this paper, we obtain an upper bound on the degree of a polynomial with real coefficients separating zeros and ones of a monotone Boolean function in the odd case of the dimension of space. Together with the previously known estimates for the even case and the lower estimate for the odd one, the final result is obtained. Similar results are obtained for the class of self-dual functions.

Keywords: monotone Boolean function, self-dual Boolean function, separating polynomial.

Vorotnikov A.S. Upper estimations of autonomous automata planar scheme

Switchable power of flat automatic schema without inputs implementation by periodical sequence is recieved in this work. A scheme is given that implements arbitrary predefined sequence of length \[ 2^n \] for positive integer \[ n \] with switchable power no more than \[ \frac{2^{n/2}}{n} \].

Keywords: boolean circuit , finite automata, models of VLSI, planar circuit, circuit power, Shannon function, upper estimations.

Demidova A.A. Analysis of cactus graphs using automata: properties and recognition time

This paper is devoted to the study of using automata with erasable colors to determine whether an arbitrary connected plane simple undirected graph is a cactus. An algorithm is given for determining this property, as well as lower and upper bounds of the number of steps that the automaton must take to complete the traversal.

Keywords: automata, graphs, cactus graphs.

Ilin I.Y. Comlexity of implementation of A-closure elementary basis in lineary automata class on finite field

In the previous work we have received compexity estimation for elementary basis implementation in the class of linear automata that preserves zero-sequence. In the current work we will find the complexity estimation for elementary basis realization by A-closure operations in linear automata under finite field [4].

Keywords: linear automata, A-closure, A-expressibility, algorithm complexity.