Serhii Lupenko
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Serhii Lupenko has a total of 65 co-authors. Here are the top co-authors by number of shared publications:
- Iaroslav Lytvynenko
Ternopil Ivan Pul’uj National Technical University
11 shared publications - Roman Butsiy
Institute of Telecommunications and Global Information Space of the National Academy of Sciences of Ukraine
6 shared publications - Andrii Horkunenko
I. Horbachevsky Ternopil State Medical University,Department of Medical Physics of Diagnostic and Therapeutic Equipment,Ternopil,Ukraine
5 shared publications - Oleksandra Orobchuk
Ternopil Ivan Puluj National Technical University
5 shared publications - Pavlo Maruschak
Department of Automation of Technological Processes and Production, Ternopil Ivan Puluj National Technical University, 46001 Ternopil, Ukraine
5 shared publications - Andrii Zozulya
Institute of Telecommunications and Global Information Space
4 shared publications
Latest Publications
- Open Access
- 11 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 571-572
AI-Supported Techniques in Academia
- 11 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 571-572
Open AccessAI-Supported Techniques in Academia
- Open Access
- 11 November 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (22) , 11175
Explainable Artificial Intelligence for Three-Hour-Ahead Natural Gas Consumption Forecasting with Uncertainty Quantification
This study investigates short-term natural gas consumption forecasting in the city of Ternopil and the contribution of different feature groups to forecasting accuracy. Synchronized natural gas consumption and meteorological observations from 2007 to 2019 were used. A total of 203 calendar, meteorological, and derived features were constructed. Six machine learning models, three naive baseline forecasts, and a seasonal ETS model were compared. Hyperparameters were optimized using Optuna, model performance was evaluated using six temporal folds with an expanding training window, and final testing was conducted on data from 2019. XGBoost achieved the lowest mean root mean squared error of 4307.13 ± 280.29 m3 across the cross-validation folds and a root mean squared error of 6517.29 m3 on the test set. No statistically significant difference was found between XGBoost and Random Forest, although both models outperformed the baseline forecasts. Feature ablation analysis showed that adding historical gas consumption features to calendar features reduced the mean root mean squared error by 81.42%. Adding meteorological features to a feature set that already included consumption history further reduced this error by 0.51%. Among the features evaluated for XGBoost, gas consumption 24 h before the target interval had the highest mean absolute SHAP value. The adaptive online conformal method produced prediction intervals that contained the observed consumption in 89.32% of cases, compared with a nominal coverage level of 90%. However, these intervals had the largest mean width among the prediction interval methods considered. The findings are limited to forecasting for a single city and one test year.
- 11 November 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (22) , 11175
Open AccessExplainable Artificial Intelligence for Three-Hour-Ahead Natural Gas Consumption Forecasting with Uncertainty Quantification
This study investigates short-term natural gas consumption forecasting in the city of Ternopil and the contribution of different feature groups to forecasting accuracy. Synchronized natural gas consumption and meteorological observations from 2007 to 2019 were used. A total of 203 calendar, meteorological, and derived features were constructed. Six machine learning models, three naive baseline forecasts, and a seasonal ETS model were compared. Hyperparameters were optimized using Optuna, model performance was evaluated using six temporal folds with an expanding training window, and final testing was conducted on data from 2019. XGBoost achieved the lowest mean root mean squared error of 4307.13 ± 280.29 m3 across the cross-validation folds and a root mean squared error of 6517.29 m3 on the test set. No statistically significant difference was found between XGBoost and Random Forest, although both models outperformed the baseline forecasts. Feature ablation analysis showed that adding historical gas consumption features to calendar features reduced the mean root mean squared error by 81.42%. Adding meteorological features to a feature set that already included consumption history further reduced this error by 0.51%. Among the features evaluated for XGBoost, gas consumption 24 h before the target interval had the highest mean absolute SHAP value. The adaptive online conformal method produced prediction intervals that contained the observed consumption in 89.32% of cases, compared with a nominal coverage level of 90%. However, these intervals had the largest mean width among the prediction interval methods considered. The findings are limited to forecasting for a single city and one test year.
- Open Access
- 1 July 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (4)
Rhythm-adaptive signal processing for effective ECG and PPG-based authentication under dynamic physiological conditions
This study presents a rhythm-adaptive framework for ECG- and PPG-based biometric authentication under mild physical activity, where heart-rate variability and cycle-to-cycle morphological changes reduce the reliability of stationary cardiac representations. The approach models ECG and PPG signals as cyclic random processes (CRPs) with variable rhythm and uses the estimated rhythm function to align cardiac cycles before extracting compact statistical features. The evaluation was performed on the walking records of the public PhysioNet Pulse Transit Time PPG Dataset (PTTPD), version 1.1.0, using all 22 subjects and the synchronized ecg and pleth_1 channels. The protocol is within-session because PTTPD provides one recording session per subject/activity; therefore, the highest ECG values are interpreted as strong within-session separability rather than as evidence of long-term biometric permanence. In the final classification tables, rhythm-adaptive averages of three cardiac cycles were used for ECG and seven cardiac cycles for PPG, following the optimization curves. For ECG, the best hybrid classifiers reached Accuracy, Balanced Accuracy and F1-score of 1.000, with FRR = 0.000, FAR = 0.000 and HTER = 0.000 for Linear SVM. For PPG, the best configuration reached Accuracy = 0.962, Balanced Accuracy = 0.961 and F1-score = 0.960 with FRR = 0.057, FAR = 0.020 and HTER = 0.039 for Nearest Neighbors. These results show that CRP-based rhythm-adaptive processing stabilizes cardiac-cycle morphology during movement and provides an interpretable alternative to purely data-driven representations for wearable and telemedicine authentication systems.
- 1 July 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (4)
Open AccessRhythm-adaptive signal processing for effective ECG and PPG-based authentication under dynamic physiological conditions
This study presents a rhythm-adaptive framework for ECG- and PPG-based biometric authentication under mild physical activity, where heart-rate variability and cycle-to-cycle morphological changes reduce the reliability of stationary cardiac representations. The approach models ECG and PPG signals as cyclic random processes (CRPs) with variable rhythm and uses the estimated rhythm function to align cardiac cycles before extracting compact statistical features. The evaluation was performed on the walking records of the public PhysioNet Pulse Transit Time PPG Dataset (PTTPD), version 1.1.0, using all 22 subjects and the synchronized ecg and pleth_1 channels. The protocol is within-session because PTTPD provides one recording session per subject/activity; therefore, the highest ECG values are interpreted as strong within-session separability rather than as evidence of long-term biometric permanence. In the final classification tables, rhythm-adaptive averages of three cardiac cycles were used for ECG and seven cardiac cycles for PPG, following the optimization curves. For ECG, the best hybrid classifiers reached Accuracy, Balanced Accuracy and F1-score of 1.000, with FRR = 0.000, FAR = 0.000 and HTER = 0.000 for Linear SVM. For PPG, the best configuration reached Accuracy = 0.962, Balanced Accuracy = 0.961 and F1-score = 0.960 with FRR = 0.057, FAR = 0.020 and HTER = 0.039 for Nearest Neighbors. These results show that CRP-based rhythm-adaptive processing stabilizes cardiac-cycle morphology during movement and provides an interpretable alternative to purely data-driven representations for wearable and telemedicine authentication systems.
- 30 June 2026
- Published by Igor Sikorsky Kyiv Polytechnic Institute in System research and information technologies
- No. 2,p. 134-152
Automated semantic ontology construction for foresight studies using large language models
Recent advances in large language models (LLMs) enable the automated discovery of semantic structures and emerging signals within text streams, offering an opportunity to redesign foresight workflows into continuous, data-driven systems. This study aims to develop and validate an automated framework for extracting, structuring, and comparing semantic ontologies using LLMs. The paralyzed approach was used for data mining from social media platforms and filtering non-domain data. The key semantic elements, goals and hypernyms corresponded, were extracted using multiple LLM configurations, with a consensus mechanism to provide semantic reliability and minimize hallucination. The extracted elements were embedded in a high-dimensional vector space, clustered iteratively using cosine similarity, and merged hierarchically. Convergence process and structural stability were analyzed using the elbow criterion and similarity metrics. The Proposed approach provides a cost-efficient alternative to traditional expert-based foresight analysis. By integrating LLM-driven semantic extraction with quantitative clustering, it enables the identification of emerging trends, weak signals, and long-term thematic structures. The results highlight the potential of LLM-based semantic modeling as a foundation for automated foresight systems.
- 30 June 2026
- Published by Igor Sikorsky Kyiv Polytechnic Institute in System research and information technologies
- No. 2,p. 134-152
Automated semantic ontology construction for foresight studies using large language models
Recent advances in large language models (LLMs) enable the automated discovery of semantic structures and emerging signals within text streams, offering an opportunity to redesign foresight workflows into continuous, data-driven systems. This study aims to develop and validate an automated framework for extracting, structuring, and comparing semantic ontologies using LLMs. The paralyzed approach was used for data mining from social media platforms and filtering non-domain data. The key semantic elements, goals and hypernyms corresponded, were extracted using multiple LLM configurations, with a consensus mechanism to provide semantic reliability and minimize hallucination. The extracted elements were embedded in a high-dimensional vector space, clustered iteratively using cosine similarity, and merged hierarchically. Convergence process and structural stability were analyzed using the elbow criterion and similarity metrics. The Proposed approach provides a cost-efficient alternative to traditional expert-based foresight analysis. By integrating LLM-driven semantic extraction with quantitative clustering, it enables the identification of emerging trends, weak signals, and long-term thematic structures. The results highlight the potential of LLM-based semantic modeling as a foundation for automated foresight systems.
- 14 May 2026
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 211, 111748
Model-based, data-driven and hybrid biometric person authentication using cardiac signals of electrical, mechanical and optical nature
- 14 May 2026
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 211, 111748
Model-based, data-driven and hybrid biometric person authentication using cardiac signals of electrical, mechanical and optical nature
- Open Access
- 19 May 2025
- Published by MDPI AG in Forecasting
- Vol. 7 (2) , 20
Stochastic Model and Rhythm-Adaptive Technologies of Statistical Analysis and Forecasting of Economic Processes with Cyclic Components
This article presents a mathematical model of cyclical economic processes, formulated as the sum of a deterministic polynomial function and a cyclic random process that simultaneously captures trend, stochasticity, cyclicity, and rhythm variability. Building on this stochastic framework, we propose rhythm-adaptive statistical techniques for estimating the probabilistic characteristics of the cyclic component; by adjusting to rhythm changes, these techniques improve estimation accuracy. We also introduce a forecasting procedure that constructs a system of rhythm-adaptive confidence intervals for future cycles. The effectiveness of the model and associated methods is demonstrated through a series of computational experiments using Federal Reserve Economic Data. Results show that the rhythm-adaptive forecasting approach achieves mean absolute errors less than half of those produced by a comparable non-adaptive method, underscoring its practical advantage for the analysis and prediction of cyclic economic phenomena.
- 19 May 2025
- Published by MDPI AG in Forecasting
- Vol. 7 (2) , 20
Open AccessStochastic Model and Rhythm-Adaptive Technologies of Statistical Analysis and Forecasting of Economic Processes with Cyclic Components
This article presents a mathematical model of cyclical economic processes, formulated as the sum of a deterministic polynomial function and a cyclic random process that simultaneously captures trend, stochasticity, cyclicity, and rhythm variability. Building on this stochastic framework, we propose rhythm-adaptive statistical techniques for estimating the probabilistic characteristics of the cyclic component; by adjusting to rhythm changes, these techniques improve estimation accuracy. We also introduce a forecasting procedure that constructs a system of rhythm-adaptive confidence intervals for future cycles. The effectiveness of the model and associated methods is demonstrated through a series of computational experiments using Federal Reserve Economic Data. Results show that the rhythm-adaptive forecasting approach achieves mean absolute errors less than half of those produced by a comparable non-adaptive method, underscoring its practical advantage for the analysis and prediction of cyclic economic phenomena.
- Open Access
- 26 December 2024
- Published by Kyiv National University of Construction and Architecture in Environmental safety and natural resources
- Vol. 52 (4) , 92-98
Improved ontological model of the knowledge base of the expert system for decision support in the field of digital processing and computer simulation of cyclic signals
The work is devoted to the improvement of the conceptual and formal-logical model of the ontology of the subject area "Modeling and Processing of Cyclic Signals". Namely, the work developed a formal-logical model of the sub-ontology of tasks, a formal-logical model of the sub-ontology of problem-solving methods in the field of modeling and processing of cyclic signals, a formal-logical model of the sub-ontology of hardware and software tools, and a formal-logical model of the sub-ontology of the results of applying mathematical models, methods and means of processing and computer simulation of cyclic signals. The improved ontology model and its implemented prototype are the main knowledge base of the onto-oriented expert decision-making support system when solving the tasks of justified selection of mathematical models and methods for evaluating characteristics, spectral analysis and computer simulation of cyclic signals within the framework of the theory of cyclic functional relations. Examples of ontology fragments developed in the Protégé environment are given. The non-contradiction of the developed ontology was confirmed by means of Protégé's automated formal and logical reasoning. The expediency of API-based integration of the developed computer ontology with a large linguistic model, namely, with ChatGPT, is substantiated.
- 26 December 2024
- Published by Kyiv National University of Construction and Architecture in Environmental safety and natural resources
- Vol. 52 (4) , 92-98
Open AccessImproved ontological model of the knowledge base of the expert system for decision support in the field of digital processing and computer simulation of cyclic signals
The work is devoted to the improvement of the conceptual and formal-logical model of the ontology of the subject area "Modeling and Processing of Cyclic Signals". Namely, the work developed a formal-logical model of the sub-ontology of tasks, a formal-logical model of the sub-ontology of problem-solving methods in the field of modeling and processing of cyclic signals, a formal-logical model of the sub-ontology of hardware and software tools, and a formal-logical model of the sub-ontology of the results of applying mathematical models, methods and means of processing and computer simulation of cyclic signals. The improved ontology model and its implemented prototype are the main knowledge base of the onto-oriented expert decision-making support system when solving the tasks of justified selection of mathematical models and methods for evaluating characteristics, spectral analysis and computer simulation of cyclic signals within the framework of the theory of cyclic functional relations. Examples of ontology fragments developed in the Protégé environment are given. The non-contradiction of the developed ontology was confirmed by means of Protégé's automated formal and logical reasoning. The expediency of API-based integration of the developed computer ontology with a large linguistic model, namely, with ChatGPT, is substantiated.
- Open Access
- 1 October 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (19) , 3084
Abstract Cyclic Functional Relation and Taxonomies of Cyclic Signals Mathematical Models: Construction, Definitions and Properties
This work is devoted to the procedure of the construction of an abstract cyclic functional relation, which summarizes and extends the known results for a cyclically correlated random process and a cyclic (cyclically distributed) random process to the case of arbitrary cyclic functional relations. Two alternative definitions of the abstract cyclic functional relation are given, and the fundamental properties of its cyclic and phase structures are presented. The theorem on the invariance of cyclicity attributes of an abstract cyclic functional relation to shifts of its argument, and which are determined by the rhythm function of this functional relation, is formulated and proved. This theorem gives the sufficient and necessary conditions that the rhythm function of an abstract cyclic functional relation must satisfy. By specifying the range of values and attributes of the cyclicity of an abstract cyclic functional relation, the definitions of important classes of cyclic functional relations are formulated. A deductive approach to building a wide system of taxonomies of classes of deterministic, stochastic, fuzzy and interval cyclic functional relations as potential mathematical models of cyclic signals is demonstrated. A comparative analysis of an abstract cyclic functional relation with the known mathematical models of cyclic signals was carried out. The results obtained in the article significantly expand and systematize the mathematical tools of the description of cyclic signals and are the basis for the development of effective model-based technologies for processing and computer simulation of signals with a cyclic space-time structure.
- 1 October 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (19) , 3084
Open AccessAbstract Cyclic Functional Relation and Taxonomies of Cyclic Signals Mathematical Models: Construction, Definitions and Properties
This work is devoted to the procedure of the construction of an abstract cyclic functional relation, which summarizes and extends the known results for a cyclically correlated random process and a cyclic (cyclically distributed) random process to the case of arbitrary cyclic functional relations. Two alternative definitions of the abstract cyclic functional relation are given, and the fundamental properties of its cyclic and phase structures are presented. The theorem on the invariance of cyclicity attributes of an abstract cyclic functional relation to shifts of its argument, and which are determined by the rhythm function of this functional relation, is formulated and proved. This theorem gives the sufficient and necessary conditions that the rhythm function of an abstract cyclic functional relation must satisfy. By specifying the range of values and attributes of the cyclicity of an abstract cyclic functional relation, the definitions of important classes of cyclic functional relations are formulated. A deductive approach to building a wide system of taxonomies of classes of deterministic, stochastic, fuzzy and interval cyclic functional relations as potential mathematical models of cyclic signals is demonstrated. A comparative analysis of an abstract cyclic functional relation with the known mathematical models of cyclic signals was carried out. The results obtained in the article significantly expand and systematize the mathematical tools of the description of cyclic signals and are the basis for the development of effective model-based technologies for processing and computer simulation of signals with a cyclic space-time structure.
- 8 May 2024
- Published by Elsevier in Digital Signal Processing
- Vol. 151
Rhythm-adaptive statistical estimation methods of probabilistic characteristics of cyclic random processes
- 8 May 2024
- Published by Elsevier in Digital Signal Processing
- Vol. 151
Rhythm-adaptive statistical estimation methods of probabilistic characteristics of cyclic random processes