A. V. Yashchenko

Kuban State Medical University

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Alexey Potapov

All-Russian Research Institute Fisheries and Oceanography

ВИ

В. И. Ванчурин

D. Mendeleyev University of Chemical Technology of Russia

Anatoly Belikov

ЕА

Е. А. Бруштейн

D. Mendeleyev University of Chemical Technology of Russia

S. A. Rodionov

King Abdullah University of Science and Technology

Maxim Peterson

ITMO University

E. V. Golovnya

Scientific Research and Design Institute of Nitric Industry and Organic Synthesis Products

O.V. Dadykina

Bryansk State Technical University

O. M. Asipovich

Kuban State Medical University

A.Yu. Goryaynova

State Clinical Hospital "Clinical Oncology Center No. 1

Cooperative Institutions

Kuban State Medical University

ITMO University

Bryansk State Technical University

State Clinical Hospital "Clinical Oncology Center No. 1

Scientific Research and Design Institute of Nitric Industry and Organic Synthesis Products

Donetsk Institute for Physics and Engineering named after O.O. Galkin

St Petersburg University

D. Mendeleyev University of Chemical Technology of Russia

National Academy of Sciences of Ukraine

Oncological Dispensary No. 2

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Distillation of neural network models for detection and description of key points of images

arXiv (Cornell University) (2020)

A. V. Yashchenko Anatoly Belikov Maxim Peterson Alexey Potapov

Image matching and classification methods, as well as synchronous location and mapping, are widely used on embedded and mobile devices. Their most resource-intensive part is the detection and description of the key points of the images. And if the classical methods of detecting and describing key points can be executed in real time on mobile devices, then for modern neural network methods with the best quality, such use is difficult. Thus, it is important to increase the speed of neural network models for the detection and description of key points. The subject of research is distillation as one of the methods for reducing neural network models. The aim of thestudy is to obtain a more compact model of detection and description of key points, as well as a description of the procedure for obtaining this model. A method for the distillation of neural networks for the task of detecting and describing key points was tested. The objective function and training parameters that provide the best results in the framework of the study are proposed. A new data set has been introduced for testing key point detection methods and a new quality indicator of the allocated key points and their corresponding local features. As a result of training in the described way, the new model, with the same number of parameters, showed greater accuracy in comparing key points than the original model. A new model with a significantly smaller number of parameters shows the accuracy of point matching close to the accuracy of the original model.

10.48550/arxiv.2006.10502

Cite

Citations (1)

Goodpoint: unsupervised learning of key point detection and description

Scientific and technical journal of information technologies mechanics and optics (2021)

Anatoly Belikov Alexey Potapov A. V. Yashchenko

Subject of Research. The paper presents the study of algorithms for key point detection and description, widely used in computer vision. Typically, the corner detector acts as a key point detector, including neural key point detectors. For some types of images obtained in medicine, the application of such detectors is problematic due to the small number of detected key points. The paper considers a problem of a neural network key point detector training on unlabeled images. Method. We proposed the definition of key points not depending on specific visual features. A method was considered for training of a neural network model meant for detecting and describing key points on unlabeled data. The application of homographic image transformation was basic to the method. The neural network model was trained to detect the same key points on pairs of noisy images related to a homographic transformation. Only positive examples were used for detector training, just points correctly matched with features produced by the neural network model for key point description. Main Results. The unsupervised learning algorithm is used to train the neural network model. For the ease of comparison, the proposed model has a similar architecture and the same number of parameters as the supervised model. Model evaluation is performed on the three different datasets: natural images, synthetic images, and retinal photographs. The proposed model shows similar results to the supervised model on the natural images and better results on retinal photographs. Improvement of results is demonstrated after additional training of the proposed model on images from the target domain. This is an advantage over a model trained on a labeled dataset. For comparison, the harmonic average of such metrics is used as: the accuracy and the depth of matching by descriptors, reproducibility of key points and image coverage. Practical Relevance. The proposed algorithm makes it possible to train the neural network key point detector together with the feature extraction model on images from the target domain without costly dataset labeling and reduce labor costs for the development of the system that uses the detector.

10.17586/2226-1494-2021-21-1-92-101

Cite

Citations (5)

Application of generative deep learning models for approximation of image distribution density

Journal of Optical Technology (2019)

A. V. Yashchenko Alexey Potapov S. A. Rodionov Innokentii Zhdanov Oleksandr Shcherbakov

Generative neural network models for visual concept learning and the problem of approximating image distribution density are studied. A criterion for an image to belong to a simulated class is introduced based on an estimate of the probability in the space of latent variables and reconstruction errors. Several generative deep learning models are compared. Quality estimates of the solution to the problem of one-class classifications for a set of images of handwritten digits are experimentally obtained.

Generative model

Density estimation

10.1364/jot.86.000769

Cite

Citations (1)

Development of an information system of geological monitoring indicators of subsoil use

Valery Dadykin O.V. Dadykina Julia Sinyukova A. V. Yashchenko Christina Pestrakova

The problems of rational use of natural resources and subsoil, in particular, are in the sphere of constant attention both from state bodies and from subsoil using enterprises. The data about the subsoil is enlarged by conducting geological exploration, subsequent interpretation of their results, submission of reports and other accompanying documents to the relevant geological information funds. However, the accumulation of primary and interpreted geological information does not contribute to the formation of a unified knowledge base that could allow to prepare analytical information automatically to support management decision-making. To eliminate this problem, it is advisable to use an ontological approach. In this case it is connected with the construction of cause-effect relations between elements (entities) and filling the resulting ontological model with factual information about the geological indicators obtained as a result of the work carried out. The task of using automated information systems to assess the efficiency of subsoil areas is also becoming very relevant. This is due to the increasing role of expert information and analytical systems in the structure of software products used for analyzing the results of geological exploration and determining new priorities in the research of mineral resources of solid minerals and groundwater for the needs of the region. The decrease in the number of geologists in this field leads to the need to design a system of geological assessment of the prospects of subsoil areas based on an ontological approach. The application of ontology is aimed to reproduce the chain of decision-making by an expert, but in a new view based on associative relations and empirical rules.

Subsoil

10.1117/12.3024921

Cite

Citations (0)

Method of instance normalization in deep-learning-based models for re-identification

Journal of Optical Technology (2020)

A. V. Yashchenko S. A. Rodionov Alexey Potapov

In this work, models for re-identification of pedestrians, based on deep learning, are studied. The models considered here use only images of pedestrians and do not require additional spatiotemporal information that can vary in different instances. A modification of the investigated models is proposed, based on the normalization of feature maps. In the best case, an assessment of the accuracy of the model demonstrates a 10% improvement over unmodified models.

Normalization

Identification

10.1364/jot.87.000487

Cite

Citations (0)

Simulation of the ammonia oxidation process over an oxide monolith catalyst

Catalysis in Industry (2010)

В. С. Бесков E. A. Brunshtein В. И. Ванчурин E. V. Golovnya A. V. Yashchenko

Monolith

Honeycomb

Nitric acid

Oxidation process

Catalytic Oxidation

Honeycomb structure

10.1134/s2070050410030104

Cite

Citations (7)

Assessment of the Nutritional Status and Skeletal Muscle Condition in Patients With Resectable Gastric Cancer

Innovative medicine of Kuban (2024)

И. Б. Уваров O. M. Asipovich S. N. Derbenev A. V. Yashchenko D. V. Andreev

Background: Patients with gastric cancer (GC) are at high risk of nutrient deficiencies (ND) that can negatively affect the postoperative period and long-term treatment outcomes. Objective: To assess the nutritional status (NS) and skeletal muscle condition in patients with resectable stage I-III GC. Materials and methods: Our study included 102 GC patients: 64 men (62.7%) and 38 women (37.3%) aged 31 to 77 years. During the assessment we used the NRS-2002 system, determined laboratory markers of the NS, calculated the Nutritional Risk Index (NRI), and evaluated the condition of adipose tissue and skeletal muscles using computed tomography. Results: We found that 70.6% of the patients had ND: NRS-2002 score < 3 points (52%) and NRS-2002 score ≥ 3 points (18.6%). Risk factors for ND were tumor localization in the cardia and stage III GC. Laboratory markers of ND were transferrin and NRI. Among the examined patients (n = 80) sarcopenia was detected in 10 (12.5%) patients: 1 patient (1/25; 4.0%) from the group without ND, 6 patients (6/42; 14.3%) from the group with NRS-2002 score < 3 points, and 3 patients (3/13; 23.1%) from the group with NRS2002 score ≥ 3 points, ( P = .212). In multivariate logistic regression analysis, independent factors associated with sarcopenia were age, body mass index, and total protein. Conclusions: All GC patients selected for radical surgery should undergo nutritional assessment using special tools (NRS-2002, NRI) and evaluation of the skeletal muscle condition. Sarcopenia may occur in patients with no clinical or laboratory evidence of ND. In GC patients at nutritional risk, preand postoperative nutritional support should be considered to reduce the ND-related risks.

10.35401/2541-9897-2024-9-1-69-77

Cite

Citations (0)

Structure and Properties of А356—AlCuFe Composite Obtained Using Electromagnetic Agitation

METALLOFIZIKA I NOVEISHIE TEKHNOLOGII (2016)

S.S. Polishchuk A.L. Berezina А. А. Давиденко V. Z. Spuskanyuk V. N. Fixsen

10.15407/mfint.36.09.1189

Cite

Citations (0)

PAYMENT FACTOR FOR POWER IN THE POWER INDUSTRY: PLUSES AND MINUSES FOR THE ECONOMY

Economics Profession Business (2022)

I. S. Mezhov A. V. Yashchenko

The article deals with the problem of reproduction of generating capacities in the aspect of building a rational investment mechanism. Since it is now working, it cannot satisfy consumers of electricity. Almost all experts say that there is only one alternative — reducing the cost of energy consumption in order to increase the competitiveness of the economy. The pricing mechanism does not allow generating companies to generate capital for their own development. A method was devised to withdraw funds from consumers to support and commission generating capacities through the so-called capacity sale mechanism. However, no matter how this non-economic method of withdrawing funds from consumers is argued, in the end, these amounts place an additional burden on the economy. Obviously, there cannot be an efficiently functioning electric power industry without the «consumption» sector, without developing and growing industry and, as a result, growing productive power consumption. One of the conditions for economic growth is the condition of economically justified production costs. Due to geographic and country conditions, the cost of electrical energy for industrial consumers in the Russian Federation is higher than abroad, and consumption due to climatic conditions is higher. In terms of electricity costs, our industry is losing the competition.

Consumption

Investment

Electric Power Industry

10.14258/epb202222

Cite

Citations (0)

Changes in the N2O concentration in the gas tract of a UKL-7 unit in the production of nonconcentrated nitric acid

Catalysis in Industry (2012)

Е. А. Бруштейн В. И. Ванчурин A. V. Yashchenko

Nitrous acid

Nitrous oxide

Nitric acid

10.1134/s2070050412040058

Cite

Citations (0)