Self-organizing Kohonen Research Articles

РЕСУРСНЫЙ ПОТЕНЦИАЛ СЕЛЬСКОХОЗЯЙСТВЕННЫХ ОРГАНИЗАЦИЙ КАК КОНКУРЕНТНОЕ ПРЕИМУЩЕСТВО ЭФФЕКТИВНОГО РАЗВИТИЯ СЕЛЬСКОГО ХОЗЯЙСТВА

The resource potential of agricultural organizations is a complex system that combines natural, land, material, technical, and labor resources in an optimal ratio in order to increase the efficiency of agricultural production, which determines the competitive advantages of organizations in the agro-food market. The assessment of the competitive capabilities of agricultural organizations in Ulyanovsk region depending on the availability of resources was carried out on the basis of calculations of interrelated indicators for two blocks: operational (characterizes the components of the resource potential) and investment (reflects the conditions for the formation and use of the resource potential). The analysis of the calculated data on the sum of places using the visualization tool “box-and-whisker plot” allows us to note the increase in the importance of the operating unit in 2022 compared to 2017. The highest financial results were achieved by agricultural organizations, which are characterized by a more uniform formation of revenue from commodity resources of crop and livestock production, are better provided with fixed and working capital, which gives them competitive advantages in the food market. Cluster analysis by visualizing the obtained results obtained using self-organizing Kohonen maps on the analytical platform Deductor made it possible to identify four clusters based on a set of indicators characterizing the composition and efficiency of using resource potential. An integrated approach to characterizing the resource potential makes it possible to justify the priority directions of its balanced development, taking into account the existing competitive advantages.

Achieving the Best Symmetry by Finding the Optimal Clustering Filters for Specific Lighting Conditions

This article explores the efficiency of various clustering methods for image segmentation under different luminosity conditions. Image segmentation plays a crucial role in computer vision applications, and clustering algorithms are commonly used for this purpose. The search for an adaptive clustering mechanism aims to ensure the maximum symmetry of real objects with objects/segments in their digital representations. However, clustering method performances can fluctuate with varying lighting conditions during image capture. Therefore, we assess the performance of several clustering algorithms—including K-Means, K-Medoids, Fuzzy C-Means, Possibilistic C-Means, Gustafson–Kessel, Entropy-based Fuzzy, Ridler–Calvard, Kohonen Self-Organizing Maps and MeanShift—across images captured under different illumination conditions. Additionally, we develop an adaptive image segmentation system utilizing empirical data. Conducted experiments highlight varied performances among clustering methods under different luminosity conditions. This research enhances a better understanding of luminosity’s impact on image segmentation and aids the method selection for diverse lighting scenarios.

Typology of professional trajectories of gifted individuals using neural network analysis

Based on a data set (100 biographies) created by the authors through content analysis of biographical material about outstanding scientists of the 19th and 20th centuries in the humanities and natural sciences, the clustering of professional trajectories of gifted individuals was carried out. Neural network analysis based on self-organizing Kohonen maps was used as a clustering method. The professional trajectories were formed within the framework of the behavioral model of the linear-stage approach to studying life cycles. Within this approach, career and professional self-realization are understood as a sequence of evolutionary stages fixed in their order of occurrence. Each stage was encoded, and the biographies were transformed into a vector system. In turn, the task of clustering consisted in dividing a hundred vectors into typical groups with several real-valued coordinates. The criteria for the quality of clustering were the minimum sum of quantization errors and the silhouette coefficient. As a result of the study, seven professional trajectories of gifted individuals were identified and interpreted. The analysis of trajectories was carried out from the point of view of the speed of success (average age of success) and those factors and conditions of the life path that could affect either rapid or slow achievement of professional goals and self-realization. This example demonstrates the possibilities and limitations of using neural network analysis for solving similar research tasks, especially when working with complex cluster forms and finding their optimal number.

Credit Risk Assessment Using a Combined Approach of Supervised and Unsupervised Learning

In the financial industry, credit scoring is a crucial tool for assessing credit risk. The study aims to enhance the accuracy and reliability of credit scoring by combining supervised and unsupervised learning methods. We propose an integrated model that combines Kohonen's Self-Organizing Maps (SOM) with the Random Forest algorithm to provide a more comprehensive analysis of credit card user data. Key features for model training were identified through feature selection and extraction. Experimental results show that the integrated model improved the AUC from 0.82 to 0.89, increased user satisfaction from a score of 3.8 to 4.35, and boosted usage rates by 12.5%. Additionally, the integrated model significantly enhanced the discrimination and prediction accuracy of user credit risk. These findings indicate that the combination of unsupervised learning with Kohonen's Self-Organizing Maps and supervised learning with Random Forest can effectively improve the accuracy of credit scoring, providing financial institutions with a more reliable tool for credit risk assessment.

Modelling of technical, environmental, and economic evaluations of the effect of the organic loading rate in semi-continuous anaerobic digestion of pre-treated organic fraction municipal solid waste

The study concerned technical feasibility, economic profitability, and carbon footprint (CF) analysis of semi-continuous anaerobic digestion (sAD) of organic fraction of municipal solid waste (OFMSW). The research assessed the pre-treatment effect on sAD by varying organic loading rates (OLR) from 3.38 to 6.75 kgvs/m3d. Three sAD configurations were investigated: hydrodynamic-cavitated (HC-OFMSW), enzymatically pre-treated (EN-OFMSW), and non-pre-treated (AD-OFMSW). Principal Component Analysis and Supervised Kohonen's Self-Organizing Maps combined the experimental, economic, and environmental evaluations.The sAD configurations were grouped predominantly according to the OLR however, within each OLR group the configurations were clustered according to the pre-treatments.The finding highlighted that pre-treatments offset inhibition in sAD of OFMSW due to the OLR increase, being economically profitable and CF negative up to 4.50 kgvs/m3d for EN-OFMSW and to 5.40 kgvs/m3d for HC-OFMSW. Whereas sAD-OFMSW remained economically and environmentally viable only up to 3.87 kgvs/m3d.HC-OFMSW reached the highest performance. In detail, for HC-OFMSW the NPV and CF ranged from 17679.30 to 43827.12 euros and from −51.08 to −407.210 kg CO2eq/1 MWh daily produced, by decreasing the OLR from 5.40 to 3.87 kgvs/m3d.These results are fundamental since pre-treatment is usually expensive due to additional energy or chemical requirements.

СТРУКТУРНО-ФУНКЦІОНАЛЬНИЙ РОЗПОДІЛ РИНКУ БАНКІВСЬКИХ ПОСЛУГ УКРАЇНИ

The article compares the structure of the banking services market as of January 1, 2022 and September 1, 2024. During this period, banks were under the influence of harsh external factors of the full-scale war in Ukraine. The number of banks decreased from 71 to 62, there was a redistribution of attracted and placed resources, a change in the amount of assets and business models of banks in the market. The issue of market redistribution and specialization of banks is an important scientific and practical problem. The National Bank of Ukraine uses the concept of the business model of financial asset management to assess the place of a particular bank in the market. The bank's business model indicates the bank's approach to managing financial assets in order to generate cash flows. Comparison of business models is currently not used to assess the state of the banking system. Evaluating the structure and state of the banking system as of the reporting dates allows identifying changes in its financial stability and risk profile, adapting regulatory and supervisory measures to specific conditions. The article uses the method of structural and functional groups of banks (SFGB), which combines homogeneous banks according to the indicators of the structure of assets, liabilities, income, expenses and other indicators of the quality of assets, efficiency of banks. The groups are formed using Kohonen's self-organization maps according to the data of the banks' published reports. . Comparing the business models of different banks, combining them into homogeneous groups and evaluating the market ratio by the number of banks in each group allows us to describe the market structure. For research, it is important to estimate both the number of banks and the total size of their assets. Each group has its own risk profile. The ratio of groups in the system characterizes the state of its financial stability. The SFGB method makes it possible to assess the overall financial condition of the system, to reveal the hidden properties of homogeneous groups of banks. The study showed qualitative changes in the characteristics of individual groups, their causes and impact on the system's stability.

Neural networks for bioreactor control solutions

The use of machine learning has the potential to improve the control of bioreactors. The aim of this research was to use self-organising Kohonen maps based on algorithms built from the composition of the taxonomic structure of the bioreactor. By adjusting the weights of the map neurons, we can infer internal unobservable dependencies in the input data structures based on the results. Using our chosen model, we will gain a deeper understanding of the taxonomic composition of the bacterial community, which will allow us to better manage fermentation processes in bioreactors.

Modeling student clusters on the basis of self-organizing Kohonen charts

Modeling student clusters on the basis of self-organizing Kohonen charts

Нейронные сети в исследовании человеческого капитала как источника экономической безопасности Российской Федерации

According to the national goal of preserving the health and well-being of the population of the Russian Federation, neural network modeling of the dynamics of the current state of organizational and economic efficiency of acyclic sports certifying human capital, which is one of the significant foundations of innovative development and guarantees of national, including economic, security of the state, was carried out. The data of the Ministry of Sports of the Russian Federation for the period 2014—2020 are considered and investigated. on the development of acyclic sports in order to improve the efficiency of organizational and economic activities in the sports field, increase the life expectancy of the Russian population, increase human capital and support the socio-economic growth of the Russian Federation. Cluster analysis was performed using a new promising methodological approach — neural networks, which make up one of the sections of artificial intelligence. Data clustering was carried out on the basis of neural networks — self-organizing Kohonen maps for 6 indicators certifying the evolution of acyclic sports, using information technology. The features of the development of acyclic sports as one of the components of the process of human capital formation, which were also influenced by the great challenges of the modern stage of civilization development, are determined. The different level of organizational and economic activity in the sphere of sports reserve training allowed us to obtain a ranking of acyclic sports in four clusters. The composition and characteristics of each cluster are presented. The conducted research using neural network technologies made it possible to assess the development of the sport under study in the Russian Federation. The results of the study contain practical significance in the implementation of a comparative analysis of the development of acyclic sports. They can be taken into account in the strategic planning of the development of the sports sphere in the context of an increase in the quality of human capital related to the dominant internal factors of the socio-economic potential of the Russian Federation.

Study of Thermal Aging of the PCM Using Acoustic Emission and Optical Microscopy

This work presents results of a study of the influence assessment based on the recorded acoustic emission (AE) parameters of thermo-oxidative aging conditions on the destruction process of a polymer composite material (PCM). The objects of the study were specimens cut from a fiberglass reinforced plastic (FGRP) plate. The plate was made by vacuum infusion technique using Derakane 411-350 resin and 9 layers of St-62004 glass fabric. Specimens aging were done by holding in a muffle furnace for 96 hours at temperatures of 60, 100, 120, and 200 ° C. Mechanical test was the three-point static bending method. For the AE recording was used a hardware-software complex developed at KnASU. The AE signal Fourier spectra were two-stage clustered with the self-organizing Kohonen map according to the technique previously developed and tested by the authors. The types of the PCM structure damage were characterized by the obtained clusters centroids. The fracture process kinetics is described depending on the conditions of thermo-oxidative aging and based on the accumulation of clusters during mechanical tests. The negative influence of high temperatures on the polymer matrix degradation, leading to a decrease in the ability of the matrix to effectively distribute internal stresses over the PCM volume due to the adhesion corruption with the reinforcing material, has been established.

Fault Detection and Diagnosis of a 3-Phase Induction Motor Using Kohonen Self-Organising Map

This paper uses the Kohonen Self-Organising Map (KSOM) to detect, diagnose, and classify induction motor faults. A series of simulations using models of the 3-phase induction motor based on real industrial motor parameters were performed using MATLAB/Simulink under fault conditions such as inter-turn, power frequency variation, over-voltage and unbalance in supply voltage. The model was trained using the input signals of the various fault conditions. Various faults from an unseen induction motor were fed to the model to test the model’s ability to detect and classify induction motor faults. The KSOM adapted to the conditions of the unseen motor, detected, diagnosed and classified these faults with an accuracy of 94.12%.

Identification of pathobiomechanical markers of statokinesiograms on the example of neural network identification of a post-stroke state

The purpose of this study is neural network modeling and determination of the parameters of statokinesiograms, which are carriers of useful information about the features of postural regulation, which determined the obtained trajectory of movements of the human center of mass. A technique for obtaining informative markers by identifying clustering centroids based on self-organizing Kohonen neural networks with the Euclidean metric has been developed. Kohonen networks trained without a teacher (that is, without the use of a priori diagnostic information about the state of the subjects) are a powerful and informative method that allows you to obtain not only graphic, but also mathematical markers of health disorders in subjects, to explore the biomechanics of micromovements of the center of pressure based on a flexible mathematical apparatus – neural network cluster analysis. On the example of the identification of the post-stroke state, a neural network analysis of stabilometric data was carried out and a method for identifying pathobiomechanical markers of statokinesiograms was shown, which made it possible to standardize the type affiliation of the groups of subjects. With the help of neural networks, it was possible to identify clusters that are amenable to biomedical interpretation with a reliability of up to 95.9 %, which determines the theoretical significance of the results obtained. Decision trees and a supervised multilayer neural network were also considered. A multilayer neural net-work made it possible to identify markers of health disorders with a probability of 71.9 %.

Diagnosis of Stator Winding and Permanent Magnet Faults of PMSM Drive Using Shallow Neural Networks

This paper presents the application of shallow neural networks (SNNs): multi-layer perceptron (MLP) and self-organizing Kohonen maps (SOMs) to the early detection and classification of the stator and rotor faults in permanent magnet synchronous motors (PMSMs). The neural networks were trained based on the vector coming from measurements on a real object. The elements of the input vector of SNNs constituted the selected amplitudes of the diagnostic signal spectrum. The stator current and axial flux were used as diagnostic signals. The test object was a 2.5 kW PMSM motor supplied by a frequency converter operating in a closed-loop control structure. The experimental verification of the proposed diagnostic system was carried out for variable load conditions and values of the supply voltage frequency. The obtained results were compared with an approach based on a deep neural network (DNN). The research presented in the article confirm the possibility of detection and assessing the individual damage of stator winding and permanent magnets as well as the simultaneous faults of the PMSM stator and rotor using SNNs with simple signal preprocessing.

Progression of Selected Parameters of the Clinical Profile of Patients with Periodontitis Using Kohonen's Self-Organizing Maps.

(1) Background: Periodontitis is an inflammatory condition that affects the tissues surrounding the tooth and causes clinical attachment loss, which is the loss of periodontal attachment (CAL). Periodontitis can advance in various ways, with some patients experiencing severe periodontitis in a short period of time while others may experience mild periodontitis for the rest of their lives. In this study, we have used an alternative methodology to conventional statistics, self-organizing maps (SOM), to group the clinical profiles of patients with periodontitis. (2) Methods: To predict the periodontitis progression and to choose the best treatment plan, we can use artificial intelligence, more precisely Kohonen's self-organizing maps (SOM). In this study, 110 patients, both genders, between the ages of 30 and 60, were included in this retrospective analysis. (3) Results: To discover the pattern of patients according to the periodontitis grade and stage, we grouped the neurons together to form three clusters: Group 1 was made up of neurons 12 and 16 that represented a percentage of slow progression of almost 75%; Group 2 was made up of neurons 3, 4, 6, 7, 11, and 14 in which the percentage of moderate progression was almost 65%; and Group 3 was made up of neurons 1, 2, 5, 8, 9, 10, 13, and 15 that represented a percentage of rapid progression of almost 60%. There were statistically significant differences in the approximate plaque index (API), and bleeding on probing (BoP) versus groups (p < 0.0001). The post-hoc tests showed that API, BoP, pocket depth (PD), and CAL values were significantly lower in Group 1 relative to Group 2 (p < 0.05) and Group 3 (p < 0.05). A detailed statistical analysis showed that the PD value was significantly lower in Group 1 relative to Group 2 (p = 0.0001). Furthermore, the PD was significantly higher in Group 3 relative to Group 2 (p = 0.0068). There was a statistically significant CAL difference between Group 1 relative to Group 2 (p = 0.0370). (4) Conclusions: Self-organizing maps, in contrast to conventional statistics, allow us to view the issue of periodontitis advancement by illuminating how the variables are organized in one or the other of the various suppositions.

Finding dependencies in the corporate environment using data mining

The article analyses the influence of factors of the work environment, as well as the non-work environment, on the employee's departure from the company. A dataset containing 1470 data rows with 14 attributes belonging to the company's employees was selected for the analysis. The method of self-organising Kohonen maps was used, which allow to study the structure of the data and identify hidden patterns, as well as the method of artificial neural networks, which allow to analyse large amounts of data and find hidden relationships that may not be obvious to humans. In the course of the work, the errors of the methods were determined, several experiments with different number of factors were conducted, and the dependence between the number of factors and the magnitude of the error of the algorithms was revealed. For both methods and each experiment, conjugacy tables were obtained, which contain the classification results obtained by the methods. In addition, a correlation analysis was performed to determine the degree of association between the factors and the target variable.

Analysis of human capital development in Russia by means of physical culture and sports using neural network modeling

Subject. The article investigates the human capital in the regions of the Russian Federation in the context of health promotion and active longevity of the population of the country. Objectives. The purpose is to examine the state of human capital driven by the development of physical culture and sports in Russian regions, using neural network modeling. Methods. Neural networks are used as a method of studying the multidimensional statistical data. The method of data clustering based on neural network modeling is not affiliated with model constraints. This method is aimed at self–training of neural networks, i.e. self-organizing Kohonen maps, and is a promising means of visual representation of multidimensional data space. Results. Using the neural network modeling, we placed 85 Russian regions in six cluster formations. We assessed the influence of each studied indicator on cluster construction. The paper demonstrates a strong difference in the number of regions of the Russian Federation in clusters, presents average values of the considered indicators in clusters for 2021. Conclusions. The study shows uneven development of human capital in Russian regions from the point of view of physical culture and sports, which predetermines different strategies for the development of the regional sports sphere. The positive trend in the number of sports facilities, personnel, and population engaged in physical culture and sports is facilitated by a systems approach based on program documents, which focuses on strengthening the health of citizens and their active longevity.

New Type of Aeronautical Risk Assessment: Performance of Kohonen SelfOrganizing Maps in Identifying Brazilian Aircraft with Greater Associated Risks

The purpose of this paper is to present a new way of assessing aeronautical risk using a configuration of Kohonen Self-Organizing Maps (SOM) to identify the Brazilian aircrafts more likely to be involved in aeronautical accidents and the riskiest Brazilian aircrafts. The technique described is classified as predictive for managing aeronautical risks, according to DOC 9859, and can be used both in the context of prevention and investigation of aeronautical accidents/incidents, as well as in the context of the insurance industry. Using this technique, it was possible to identify the 147 Brazilian aircraft with the highest associated probabilities of occurrence of aeronautical accidents, and the 180 with the highest associated risks. Five years after this identification, the respective percentages of aeronautical accidents/incidents were 34% and 27%. The application of this technique can help achieve the objective of the aeronautical community in determining what, where, and when the next aeronautical accidents and/or incidents will occur. Another aspect of the present work is to demonstrate that data collected by the national civil aviation agency in Brazil can be used to implement a predictive methodology for the management of safety in civil aviation.

Implementasi Jaringan Saraf Tiruan Pengenalan Pola Aksara Batak Simalungun Menggunakan Kohonen Self Organizing Map

The Simalungun Batak script contained in the Batak Letter consists of several variants of forms depending on the language and region, in general there are five variants of the Batak Letter in Sumatra, namely Karo, Toba, Pakpak, Simalungun, Angkola Mandailing. This script is not widely known by the datu, namely people who are respected by the Batak community for mastering magic, fortune-telling, and calendaring so that errors still often occur in the introduction of this script, even though this script can be found in various libraries, namely the traditional books of the Batak community. By utilizing Artificial Neural Networks and using the Kohonen Self Organizing Map method which can be implemented in the Simalungun Batak Letter Recognition Application so that it can make it easier for the public to recognize the Simalungun Batak script. The results of the study that applied an artificial neural network with the kohonen self-organizing map algorithm made the recognition of the Simalungun Batak letter pattern easy for the public to recognize and could also preserve one of the cultures, namely the Simalungun Batak script and make it easier for users to implement it into the android-based application.

Deep Learning Mechanism for Predicting the Axillary Lymph Node Metastasis in Patients with Primary Breast Cancer.

The second largest cause of mortality worldwide is breast cancer, and it mostly occurs in women. Early diagnosis has improved further treatments and reduced the level of mortality. A unique deep learning algorithm is presented for predicting breast cancer in its early stages. This method utilizes numerous layers to retrieve significantly greater amounts of information from the source inputs. It could perform automatic quantitative evaluation of complicated image properties in the medical field and give greater precision and reliability during the diagnosis. The dataset of axillary lymph nodes from the breast cancer patients was collected from Erasmus Medical Center. A total of 1050 images were studied from the 850 patients during the years 2018 to 2021. For the independent test, data samples were collected for 100 images from 95 patients at national cancer institute. The existence of axillary lymph nodes was confirmed by pathologic examination. The feed forward, radial basis function, and Kohonen self-organizing are the artificial neural networks (ANNs) which are used to train 84% of the Erasmus Medical Center dataset and test the remaining 16% of the independent dataset. The proposed model performance was determined in terms of accuracy (Ac), sensitivity (Sn), specificity (Sf), and the outcome of the receiver operating curve (Roc), which was compared to the other four radiologists' mechanism. The result of the study shows that the proposed mechanism achieves 95% sensitivity, 96% specificity, and 98% accuracy, which is higher than the radiologists' models (90% sensitivity, 92% specificity, and 94% accuracy). Deep learning algorithms could accurately predict the clinical negativity of axillary lymph node metastases by utilizing images of initial breast cancer patients. This method provides an earlier diagnostic technique for axillary lymph node metastases in patients with medically negative changes in axillary lymph nodes.

Comparison of supervised and unsupervised anomaly detection in Belle II pixel detector data

Machine learning has become a popular instrument for the search of undiscovered particles and mechanisms at particle collider experiments. It enables the investigation of large datasets and is therefore suitable to operate directly on minimally-processed data coming from the detector instead of reconstructed objects. Here, we study patterns of raw pixel hits recorded by the Belle II pixel detector, that is operational since 2019 and presently features 4 M pixels and trigger rates up to 5 kHz. In particular, we focus on unsupervised techniques that operate without the need for a theoretical model. These model-agnostic approaches allow for an unbiased exploration of data while filtering out anomalous detector signatures that could hint at new physics scenarios. We present the identification of hypothetical magnetic monopoles against Belle II beam background using self-organizing kohonen maps and autoencoders. These two unsupervised algorithms are compared to a Multilayer Perceptron and a superior signal efficiency of the Autoencoder is found at high background-rejection levels. Our results strengthen the case for using unsupervised machine learning techniques to complement traditional search strategies at particle colliders and pave the way to potential online applications of the algorithms in the near future.