Bulletin of Taras Shevchenko National University of Kyiv. Series: Physics and Mathematics
Latest Publications


TOTAL DOCUMENTS

184
(FIVE YEARS 184)

H-INDEX

1
(FIVE YEARS 1)

Published By Taras Shevchenko National University Of Kyiv

2218-2055, 1812-5409
Updated Wednesday, 07 July 2021

Author(s):  
A. S. Dzhoha

Online learning under delayed feedback has been recently gaining increasing attention. Learning with delays is more natural in most practical applications since the feedback from the environment is not immediate. For example, the response to a drug in clinical trials could take a while. In this paper, we study the multi-armed bandit problem with Bernoulli distribution in the environment with delays by evaluating the Explore-First algorithm. We obtain the upper bounds of the algorithm, the theoretical results are applied to develop the software framework for conducting numerical experiments.


Author(s):  
A. G. Rudnitskii ◽  
M. A. Rudnytska ◽  
L. V. Tkachenko ◽  
E. D. Pechuk

Denoising is an important step in the early stage of signal preprocessing in optoacoustic applications. The efficiency of such modern noise removal methods as wavelet or curvlet filtering depends significantly on the numerical combinations and forms of wavelet transform parameters, and the multidimensional extension of such filters is rather non-trivial. These issues are serious obstacle for using of these highly effective filters in the tasks of optoacoustic reconstruction, especially in real laboratorial or medical practice. The objective of our study was to find the optimal filter, convenient for use in laboratorian and medical practice, when the types of noise are a priori unknown, and the filter settings should not take much time. In the offered work spatial filters which have only one parameter of adjustment - the size of a window are considered. Three-dimensional extensions of such well-established denoising techniques, as mean filter, median filter, their adaptive variants (Wiener spatial filter and modified median filter), as well as iterative truncated arithmetic mean filter were analyzed. The proposed filters were tested on a test set that contains versions of Shepp-Logan's three-dimensional phantom with mixtures of Gaussian and alpha-stable noise, as well as speckle noise. The identification of the best filter for simultaneous suppression of these types of interference was carried out using the theory of fuzzy sets. In our tests, a modified median filter and an iterative truncated arithmetic mean filter were rated as the best choice when the goal is to minimize aberrations when noise is not known a priory.


Author(s):  
O. H. Skurzhanskyi ◽  
A. A. Marchenko

The article is devoted to the review of conditional test generation, one of the most promising fields of natural language processing and artificial intelligence. Specifically, we explore monolingual local sequence transduction tasks: paraphrase generation, grammatical and spelling errors correction, text simplification. To give a better understanding of the considered tasks, we show examples of good rewrites. Then we take a deep look at such key aspects as publicly available datasets with the splits (training, validation, and testing), quality metrics for proper evaluation, and modern solutions based primarily on modern neural networks. For each task, we analyze its main characteristics and how they influence the state-of-the-art models. Eventually, we investigate the most significant shared features for the whole group of tasks in general and for approaches that provide solutions for them.


Author(s):  
I. Bieda

Millions of videos are uploaded each day to Youtube and similar platforms. One of the many issues that these services face is the extraction of useful metadata. There are a lot of tasks that arise with the processing of videos. For example, putting an ad is better in the middle of a video, and as an advertiser, one would probably prefer to show the ad in between scene cuts, where it would be less intrusive. Another example is when one would like to watch only through the most interesting or important pieces of video recording. In many cases, it is better to have an automatic scene cut detection approach instead of manually labeling thousands of videos. The scene change detection can help to analyze video-stream automatically: which characters appear in which scenes, how they interact and for how long, their relations and importance, and also to track many other issues. The potential solution can rely on different facts: objects appearance, contrast or intensity changed, other colorization, background chang, and also sound changes. In this work, we propose the method for effective scene change detection, which is based on thresholding, and also fade-in/fade-out scene analysis. It uses computer vision and image analysis approaches to identify the scene cuts. Experiments demonstrate the effectiveness of the proposed scene change detection approach.


Author(s):  
T. V. Kolianova

The article considers the isolated population described by the logistic equation and studies the influence of management on the change of its number. Depending on the value of the control parameter, there are three different cases of behavior of an isolated population. In the first case, when the quota is equal to the corresponding value, depending on the initial value, the population either goes to a stationary value, or dies out. In the second case, when the quota does not exceed the established value, depending on the initial population size, the population either goes to the largest stationary point, or dies out. And in the third case, when the quota exceeds the established value, regardless of the initial population size, the population dies out. Stationary points for stability in all three cases are studied and graphs of population behavior depending on different initial conditions are presented.


Author(s):  
L. L. Omelchuk ◽  
N. G. Rusina

The article presents an analysis of the educational and professional program "Informatics" of the first (bachelor's) level of higher education in the sphere of knowledge 12 "Information Technology", specialty 122 "Computer Science", implemented at the Faculty of Computer Science and Cybernetics, Taras National University of Kyiv Shevchenko with educational and professional programs of the same level and specialty of other institutions of higher education of Ukraine in terms of program results. During the analysis, they were compared with the approved standard of the first (bachelor's) level of higher education in the specialty 122 "Computer Science". In order to analyze the authors developed a database of educational programs. The ratio of program results in different programs by common specialty is analyzed.


Author(s):  
O. I. Vasylyk ◽  
I. I. Lovytska

In the paper, we consider the problem of simulation of a strictly φ-sub-Gaussian generalized fractional Brownian motion. Simulation of random processes and fields is used in many areas of natural and social sciences. A special place is occupied by methods of simulation of the Wiener process and fractional Brownian motion, as these processes are widely used in financial and actuarial mathematics, queueing theory etc. We study some specific class of processes of generalized fractional Brownian motion and derive conditions, under which the model based on a series representation approximates a strictly φ-sub-Gaussian generalized fractional Brownian motion with given reliability and accuracy in the space C([0; 1]) in the case, when φ(x) = (|x|^p)/p, |x| ≥ 1, p > 1. In order to obtain these results, we use some results from the theory of φ-sub-Gaussian random processes. Necessary simulation parameters are calculated and models of sample pathes of corresponding processes are constructed for various values of the Hurst parameter H and for given reliability and accuracy using the R programming environment.


Author(s):  
A. Yavorskyi

Analysis of Electrocardiogram (ECG) signals is an important task to save and enhance human life because a major cause of death is heart disease and the consequences. In many cases, early diagnostics of such problems can save and prolong life. In this work, we develop and present an approach to the real-time detection of Atrial Fibrillation (AF) Arrhythmia, which is a common cardiac arrhythmia affecting a large number of people. Being undetected, it develops into chronic disability or even early mortality. At the same time, This disease is hard to diagnose, especially in its early stage. A real-time automatic and non-invasive effective detection is needed to help diagnose this kind of health problem early. In-time medical intervention can save human life. ECG as a record of the heart electrical activity is widely used for detecting different heart disabilities. At the same time, AF is hard to detect due to its non-regular nature, and also because the performance of detection models depends largely on the quality of data and careful feature engineering. The research is based on the dataset from PhysioNet Computing in Cardiology Challenge 2017. It contains 8528 single-lead ECG recordings of short-term heart rhythms (9-61 sec.). Our method and the trained model reach the known state-of-the-art results in this field, but, at the same time, it is much less computationally intensive, and, thus, less power consumptive to be implemented in an embedded device.


Author(s):  
V. R. Kulian ◽  
O. O. Yunkova

In article we consider a problem of optimal investment strategy by a commercial bank building. This task is actual and the development of a procedure to solve it can help in making investment banking decisions. The general formulation of the problem consists of two criteria. The first one is to maximize the expected return, and the second is to minimize the risk of the investment transaction. Mathematical formulation of the problem is considered as a problem of nonlinear programming under constraints. The procedure for solving such a two-criteria optimization problem allows to obtain many solutions, which requires further steps to make a single optimal solution. According to the algorithm proposed in the work, the problem is divided into two separate problems of single-criteria optimization. Each of these tasks allows to obtain the optimal values of the investment vector both in terms of its expected return and in terms of investment risk. Additional constraints in the mathematical formulation of the problem, make it possible to take into account factors that, from the point of view of the investor, may influence decision-making. The procedures presented in this work allow to obtain analytical representations of formulas that describe the optimal values of the investment distribution vector for both mathematical formulations of the problem.


Author(s):  
V. P. Danko ◽  
A. V. Kovalenko ◽  
R. O. Kolomiiets

The proposed work analyzes the design features of the acousto-optical deflector and filter on paratelurite. It is shown that under certain conditions the acousto-optical deflector can be used as an acousto-optical filter (as an element that performs spectral filtering of the incident light beam). The fundamental possibility of creating a monochromatic light source with a variable wavelength and a spectrum width of about 5 nm using an acousto-optical deflector as an element that adjusts the original wavelength is shown experimentally. As a broadband light source in this system, a semiconductor laser operating in subthreshold mode was used. The dependence of the output wavelength on the acoustic frequency is obtained. The comparison of experimental data with the calculated ones is given, it is shown that they have small differences.


Sign in / Sign up

Export Citation Format

Share Document