Marzhan Yesmagambetova
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Marzhan Yesmagambetova has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Т.у. Есмагамбетов
Карагандинский университет Казпотребсоюза
2 shared publications - Ainur Zhumadillayeva
Astana IT University; L.N. Gumilyov Eurasian National University
1 shared publication - Akzhibek Amirova
School of Cybersecurity, Astana IT University, Astana 010000, Kazakhstan
1 shared publication - Karlygash Alibekkyzy
D. Serikbayev East Kazakhstan Technical University
1 shared publication - Kayrat Koshekov
Department of Aviation Technique and Technologies, Civil Aviation Academy, Almaty
1 shared publication - Oksana Mulesa
Division of Mathematics, Department of Physics, Mathematics and Technologies, Faculty of Humanities and Natural Sciences, University of Prešov, 17. Novembra St. 1, 080 01 Prešov, Slovakia
1 shared publication
Latest Publications
- Open Access
- 30 September 2026
- Published by K. Zhubanov Aktobe Regional University in Научный журнал "Вестник Актюбинского регионального университета имени К. Жубанова"
- Vol. 85 (3) , 157-167
AI AND NEURAL NETWORKS IN SPORTS ANALYTICS: A COMPREHENSIVE REVIEW OF TECHNIQUES AND APPLICATIONS
The explosive growth of sports data, including event logs, tracking streams, video records, wearable signals, and biomechanics-related measurements, has fundamentally changed the way athletic performance is studied, interpreted, and optimized. What previously relied on manual observation and conventional statistics is now increasingly supported by artificial intelligence capable of extracting patterns from large, heterogeneous, and high-frequency datasets. This review aims to systematize the principal machine learning and deep learning approaches used in contemporary sports analytics, with a particular focus on three major application domains: match outcome prediction, tactical and spatiotemporal analysis, and injury risk assessment and prevention. The article synthesizes the use of widely adopted methods, ranging from regression, decision trees, random forests, support vector machines, and gradient boosting to artificial neural networks, convolutional neural networks, recurrent architectures, and graph-based models. Special attention is given to the relationship between data modality and model selection, especially in studies using tracking data, event data, physiological inputs, and biomechanical indicators. In addition to summarizing methodological trends, the review examines how AI-driven systems contribute to coaching support, player evaluation, workload monitoring and tactical decision-making. The evidence suggests that AI and neural networks have moved beyond experimental use and are becoming a central analytical infrastructure in modern sport. At the same time, persistent limitations remain, including data heterogeneity, limited interpretability, inconsistent validation protocols, and difficulties in transferring results from controlled settings to real competitive environments. Overall, this review positions AI-based sports analytics as a rapidly evolving field with strong practical value and clear relevance for the development of next-generation intelligent systems in sport.
- 30 September 2026
- Published by K. Zhubanov Aktobe Regional University in Научный журнал "Вестник Актюбинского регионального университета имени К. Жубанова"
- Vol. 85 (3) , 157-167
Open AccessAI AND NEURAL NETWORKS IN SPORTS ANALYTICS: A COMPREHENSIVE REVIEW OF TECHNIQUES AND APPLICATIONS
The explosive growth of sports data, including event logs, tracking streams, video records, wearable signals, and biomechanics-related measurements, has fundamentally changed the way athletic performance is studied, interpreted, and optimized. What previously relied on manual observation and conventional statistics is now increasingly supported by artificial intelligence capable of extracting patterns from large, heterogeneous, and high-frequency datasets. This review aims to systematize the principal machine learning and deep learning approaches used in contemporary sports analytics, with a particular focus on three major application domains: match outcome prediction, tactical and spatiotemporal analysis, and injury risk assessment and prevention. The article synthesizes the use of widely adopted methods, ranging from regression, decision trees, random forests, support vector machines, and gradient boosting to artificial neural networks, convolutional neural networks, recurrent architectures, and graph-based models. Special attention is given to the relationship between data modality and model selection, especially in studies using tracking data, event data, physiological inputs, and biomechanical indicators. In addition to summarizing methodological trends, the review examines how AI-driven systems contribute to coaching support, player evaluation, workload monitoring and tactical decision-making. The evidence suggests that AI and neural networks have moved beyond experimental use and are becoming a central analytical infrastructure in modern sport. At the same time, persistent limitations remain, including data heterogeneity, limited interpretability, inconsistent validation protocols, and difficulties in transferring results from controlled settings to real competitive environments. Overall, this review positions AI-based sports analytics as a rapidly evolving field with strong practical value and clear relevance for the development of next-generation intelligent systems in sport.
- Open Access
- 23 September 2026
- Published by al-Farabi Kazakh National University in Journal of problems in computer science and information technologies
- Vol. 4 (3) , 3-19
SELF-SUPERVISED FOUNDATION MODELS FOR REAL-TIME DRILLING OPTIMIZATION: CROSS-MODAL KNOWLEDGE TRANSFER FROM WELL LOG REPRESENTATIONS
The optimization of real-time drilling faces significant hurdles stemming from limited data availability, complex operational environments, and a lack of machine learning models that can transfer knowledge across different geological settings. Current supervised methods deliver strong results in predicting rate of penetration and detecting stuck pipe incidents, yet they demand extensive labeled datasets for each new well and fail to tap into the geological insights contained in commonly available well log measurements. We present DrillFM, a self-supervised foundation model designed for multivariate drilling time-series data that incorporates geological information through cross-modal conditioning from pretrained well log representations. Our architecture merges masked autoencoding of drilling parameters-including weight-on-bit, rotary speed, torque, standpipe pressure, and flow rate-with a cross-modal attention mechanism that draws upon formation embeddings generated by a physics-aware well log foundation model. The training employs a drilling-adaptive curriculum that escalates task complexity progressively: beginning with point-wise masking, advancing to channel-wise masking, and culminating in interval-based prediction. On the publicly available Volve field dataset plus proprietary multi-basin data, geological conditioning cuts rate of penetration prediction error by 18.3 percent relative to the supervised multi-modal baseline. Few-shot transfer to previously unseen wells achieves comparable performance using 90 percent less labeled data. In cross-basin transfer of rate-of-penetration prediction, DrillFM's performance degrades by 38.9 percent moving from Norwegian Volve to Dutch North Sea wells, 13.0 percentage points less than the supervised multi-modal baseline's 51.9 percent degradation. The model delivers inference speeds of 8.6 milliseconds per depth sample using precomputed well log embeddings; because the reported results condition on log data from the same well being predicted, whether this speed and accuracy hold under fully real-time deployment-where only offset-well or no logs would be available-has not been separately evaluated (see Limitations). Ablation experiments show that well log foundation model pretraining is the single largest contributor, producing a 17.7 percent error increase when removed; removing geological conditioning overall accounts for a 12.5 percentage point change, and the masking curriculum contributes a further 5.7 points.
- 23 September 2026
- Published by al-Farabi Kazakh National University in Journal of problems in computer science and information technologies
- Vol. 4 (3) , 3-19
Open AccessSELF-SUPERVISED FOUNDATION MODELS FOR REAL-TIME DRILLING OPTIMIZATION: CROSS-MODAL KNOWLEDGE TRANSFER FROM WELL LOG REPRESENTATIONS
The optimization of real-time drilling faces significant hurdles stemming from limited data availability, complex operational environments, and a lack of machine learning models that can transfer knowledge across different geological settings. Current supervised methods deliver strong results in predicting rate of penetration and detecting stuck pipe incidents, yet they demand extensive labeled datasets for each new well and fail to tap into the geological insights contained in commonly available well log measurements. We present DrillFM, a self-supervised foundation model designed for multivariate drilling time-series data that incorporates geological information through cross-modal conditioning from pretrained well log representations. Our architecture merges masked autoencoding of drilling parameters-including weight-on-bit, rotary speed, torque, standpipe pressure, and flow rate-with a cross-modal attention mechanism that draws upon formation embeddings generated by a physics-aware well log foundation model. The training employs a drilling-adaptive curriculum that escalates task complexity progressively: beginning with point-wise masking, advancing to channel-wise masking, and culminating in interval-based prediction. On the publicly available Volve field dataset plus proprietary multi-basin data, geological conditioning cuts rate of penetration prediction error by 18.3 percent relative to the supervised multi-modal baseline. Few-shot transfer to previously unseen wells achieves comparable performance using 90 percent less labeled data. In cross-basin transfer of rate-of-penetration prediction, DrillFM's performance degrades by 38.9 percent moving from Norwegian Volve to Dutch North Sea wells, 13.0 percentage points less than the supervised multi-modal baseline's 51.9 percent degradation. The model delivers inference speeds of 8.6 milliseconds per depth sample using precomputed well log embeddings; because the reported results condition on log data from the same well being predicted, whether this speed and accuracy hold under fully real-time deployment-where only offset-well or no logs would be available-has not been separately evaluated (see Limitations). Ablation experiments show that well log foundation model pretraining is the single largest contributor, producing a 17.7 percent error increase when removed; removing geological conditioning overall accounts for a 12.5 percentage point change, and the masking curriculum contributes a further 5.7 points.
- Open Access
- 30 June 2026
- Published by Kazakh University of Technology and Business in Вестник КазУТБ
- Vol. 2 (31)
РАЗРАБОТКА И ИССЛЕДОВАНИЕ 1D-CNN ДЛЯ ИДЕНТИФИКАЦИИ ГАЗОВЫХ СМЕСЕЙ ПО СПЕКТРАЛЬНЫМ ДАННЫМ МУЛЬТИСПЕКТРАЛЬНОГО ДАТЧИКА
В статье рассматривается задача автоматической идентификации состава газовых смесей на основе данных мультиспектрального сенсора. Анализ газовых сред является важной задачей для промышленной безопасности, экологического мониторинга, медицинской диагностики и интеллектуальных сенсорных систем. Традиционные методы газового анализа, основанные на спектроскопии, обеспечивают высокую точность измерений, однако требуют сложного оборудования и трудоемкой обработки данных. В связи с этим актуальным направлением является применение методов искусственного интеллекта для обработки спектральных сигналов. Целью данной работы является разработка и исследование модели сверточной нейронной сети для идентификации газовых смесей по спектральным данным мультиспектрального датчика AS7265x. Для проведения экспериментов была создана лабораторная установка, включающая мультиспектральный сенсор и систему контроля условий эксперимента на основе датчика температуры и давления. В качестве входных данных использовались спектральные векторы, содержащие 18 каналов интенсивности, соответствующих различным длинам волн. Для обработки спектральных данных была разработана архитектура одномерной сверточной нейронной сети (1D-CNN), способная выявлять характерные спектральные признаки различных газов. Модель была обучена и протестирована на экспериментальном наборе данных, содержащем измерения воздуха, углекислого газа и смеси кислорода с азотом. Экспериментальные результаты показали высокую точность классификации, достигшую 100% на тестовой выборке. Полученные результаты подтверждают эффективность применения методов глубокого обучения для анализа спектральных данных и демонстрируют перспективность использования компактных мультиспектральных сенсоров в интеллектуальных системах газоанализа.
- 30 June 2026
- Published by Kazakh University of Technology and Business in Вестник КазУТБ
- Vol. 2 (31)
Open AccessРАЗРАБОТКА И ИССЛЕДОВАНИЕ 1D-CNN ДЛЯ ИДЕНТИФИКАЦИИ ГАЗОВЫХ СМЕСЕЙ ПО СПЕКТРАЛЬНЫМ ДАННЫМ МУЛЬТИСПЕКТРАЛЬНОГО ДАТЧИКА
В статье рассматривается задача автоматической идентификации состава газовых смесей на основе данных мультиспектрального сенсора. Анализ газовых сред является важной задачей для промышленной безопасности, экологического мониторинга, медицинской диагностики и интеллектуальных сенсорных систем. Традиционные методы газового анализа, основанные на спектроскопии, обеспечивают высокую точность измерений, однако требуют сложного оборудования и трудоемкой обработки данных. В связи с этим актуальным направлением является применение методов искусственного интеллекта для обработки спектральных сигналов. Целью данной работы является разработка и исследование модели сверточной нейронной сети для идентификации газовых смесей по спектральным данным мультиспектрального датчика AS7265x. Для проведения экспериментов была создана лабораторная установка, включающая мультиспектральный сенсор и систему контроля условий эксперимента на основе датчика температуры и давления. В качестве входных данных использовались спектральные векторы, содержащие 18 каналов интенсивности, соответствующих различным длинам волн. Для обработки спектральных данных была разработана архитектура одномерной сверточной нейронной сети (1D-CNN), способная выявлять характерные спектральные признаки различных газов. Модель была обучена и протестирована на экспериментальном наборе данных, содержащем измерения воздуха, углекислого газа и смеси кислорода с азотом. Экспериментальные результаты показали высокую точность классификации, достигшую 100% на тестовой выборке. Полученные результаты подтверждают эффективность применения методов глубокого обучения для анализа спектральных данных и демонстрируют перспективность использования компактных мультиспектральных сенсоров в интеллектуальных системах газоанализа.
- Open Access
- 31 March 2025
- Published by Kazakh University of Technology and Business in Вестник КазУТБ
- Vol. 1 (26)
АНЫҚ ЕМЕС ЛОГИКАҒА НЕГІЗДЕЛГЕН АҚПАРАТТЫҚ ҚАУІПСІЗДІК ҚАТЕРДІ БАҒАЛАУ МОДЕЛІ
Мақалада өнеркәсіптік заттардың интернеті (IIoT- Industrial Internet of Things) ортасындағы ақпараттық қауіпсіздік мәселесі қарастырылады. IIoT - тегі ақпараттық қауіпсіздік қатерлерді бағалау бірқатар факторлармен қиындайды: жүйенің күрделілігі мен біртектілігі, жүйенің динамикасы, таратылған желілік инфрақұрылым, стандарттар мен нұсқаулықтардың болмауы, сондай-ақ қауіпсіздіктің бұзылуының жоғарылауы. Осы сынды факторларды ескере отырып, IIoT-те ақпараттық қауіпсіздік қатерлерін бағалау белгілі бір жүйе мен саланың ерекшеліктері мен талаптарына бейімделген кешенді тәсілді қажет етеді. Қатерлерді бағалаудың мамандандырылған әдістерін қолдану және жүйенің контексті мен ерекшеліктерін ескеру қажет. Анық емес жиындар теориясының математикалық аппаратына негізделген IIoT-те ақпараттық қауіпсіздік қатерлерін бағалау әдісі ұсынылған. Бұл мақалада IIoT жүйелеріне ақпараттық қауіпсіздік қатерлеріне талдау жүргізілді, оның ішінде ең маңызды критерийлер таңдалды. Шешімдердің негізінде қабылданатын ережелер кіріс параметрлері бар логикалық формулалар түрінде тұжырымдалады. Үш анық емес қорытындылар жүйесі қолданылады: біріншісі қауіптің туындау ықтималдығын бағалау үшін, екіншісі ықтимал залалды бағалау үшін және соңғысы IIoT жүйесі үшін ақпараттық қауіпсіздік қатерді бағалау үшін. Ұсынылған әдіс негізінде IIoT ортасында ақпараттық қауіпсіздік қатерлерді бағалауды есептеу мысалдары келтірілген. Ұсынылған ғылыми тәсіл IIoT жүйелерін жобалау үшін сараптамалық шешімдерді қолдау жүйелерін құру үшін негіз бола алады
- 31 March 2025
- Published by Kazakh University of Technology and Business in Вестник КазУТБ
- Vol. 1 (26)
Open AccessАНЫҚ ЕМЕС ЛОГИКАҒА НЕГІЗДЕЛГЕН АҚПАРАТТЫҚ ҚАУІПСІЗДІК ҚАТЕРДІ БАҒАЛАУ МОДЕЛІ
Мақалада өнеркәсіптік заттардың интернеті (IIoT- Industrial Internet of Things) ортасындағы ақпараттық қауіпсіздік мәселесі қарастырылады. IIoT - тегі ақпараттық қауіпсіздік қатерлерді бағалау бірқатар факторлармен қиындайды: жүйенің күрделілігі мен біртектілігі, жүйенің динамикасы, таратылған желілік инфрақұрылым, стандарттар мен нұсқаулықтардың болмауы, сондай-ақ қауіпсіздіктің бұзылуының жоғарылауы. Осы сынды факторларды ескере отырып, IIoT-те ақпараттық қауіпсіздік қатерлерін бағалау белгілі бір жүйе мен саланың ерекшеліктері мен талаптарына бейімделген кешенді тәсілді қажет етеді. Қатерлерді бағалаудың мамандандырылған әдістерін қолдану және жүйенің контексті мен ерекшеліктерін ескеру қажет. Анық емес жиындар теориясының математикалық аппаратына негізделген IIoT-те ақпараттық қауіпсіздік қатерлерін бағалау әдісі ұсынылған. Бұл мақалада IIoT жүйелеріне ақпараттық қауіпсіздік қатерлеріне талдау жүргізілді, оның ішінде ең маңызды критерийлер таңдалды. Шешімдердің негізінде қабылданатын ережелер кіріс параметрлері бар логикалық формулалар түрінде тұжырымдалады. Үш анық емес қорытындылар жүйесі қолданылады: біріншісі қауіптің туындау ықтималдығын бағалау үшін, екіншісі ықтимал залалды бағалау үшін және соңғысы IIoT жүйесі үшін ақпараттық қауіпсіздік қатерді бағалау үшін. Ұсынылған әдіс негізінде IIoT ортасында ақпараттық қауіпсіздік қатерлерді бағалауды есептеу мысалдары келтірілген. Ұсынылған ғылыми тәсіл IIoT жүйелерін жобалау үшін сараптамалық шешімдерді қолдау жүйелерін құру үшін негіз бола алады
- Open Access
- 1 December 2022
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 28 (3) , 1463-1474
Smart technologies of the risk-management and decision-making systems in a fuzzy data environment
The purpose of this article is to provide a methodology for calculating and predicting the quality of solution implementation in complicated multi-parametric organizational and technological challenges with control agent uncertainty. The article's study findings are centered on the practical application of formal methods in predicting the outcomes of control and decision-making risks under the uncertainty of model agents. The proposed mathematics and simulation applications use a multi-agent strategy to handle the general problem of assessing quality control based on "producer risk (project customer)" and "user risk." Computer experiments with simultaneous graphical visualization of the results improve the accuracy of mathematical modeling, increasing the study's effectiveness. Under the uncertainty of system agents, a simulation model has been designed to analyze and anticipate the dependability of control and the hazards of decision-making. The suggested model is unique in that it takes into account the statistical nature of normative values as well as the rules of equal probability. To handle a frequent problem, the proposed system technique employs a dual approach. It accomplishes this by assessing the quality of the control process based on the magnitude of the risks in the decision-making system.
- 1 December 2022
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 28 (3) , 1463-1474
Open AccessSmart technologies of the risk-management and decision-making systems in a fuzzy data environment
The purpose of this article is to provide a methodology for calculating and predicting the quality of solution implementation in complicated multi-parametric organizational and technological challenges with control agent uncertainty. The article's study findings are centered on the practical application of formal methods in predicting the outcomes of control and decision-making risks under the uncertainty of model agents. The proposed mathematics and simulation applications use a multi-agent strategy to handle the general problem of assessing quality control based on "producer risk (project customer)" and "user risk." Computer experiments with simultaneous graphical visualization of the results improve the accuracy of mathematical modeling, increasing the study's effectiveness. Under the uncertainty of system agents, a simulation model has been designed to analyze and anticipate the dependability of control and the hazards of decision-making. The suggested model is unique in that it takes into account the statistical nature of normative values as well as the rules of equal probability. To handle a frequent problem, the proposed system technique employs a dual approach. It accomplishes this by assessing the quality of the control process based on the magnitude of the risks in the decision-making system.
- Open Access
- 30 March 2022
- Published by Astana IT University in Scientific Journal of Astana IT University
- No. 9,p. 61-74
INTELLECTUAL INFORMATION TECHNOLOGIES IN THE ACTIVITIES OF THE EMERGENCY RESCUE SERVICE
The paper considers the features of the use of intelligent information technologiesin the process of the emergency rescue service of the enterprise. To improve the efficiencyof the rescue services, it is necessary to carry out strategic planning and management,which would be designed to prevent the occurrence of an emergency (emergency). Effectivesubsystems of emergency proactive planning should not only predict the occurrence ofpossible emergencies, but also provide for appropriate preventive measures, and emphasisshould be placed on eliminating the underlying causes, not the emerging consequences. Thechange of modes in the event of an emergency can be effectively implemented through thedeployment of a technologically secured situational center. Decisions in emergency situationsare made in various operational situations, including crisis, and in extremely limited time.Nevertheless, they must be taken in a timely manner, be as reasonable as possible, and ensurethe fullest and most effective use of available opportunities. Decision-making processescan be based on the personal experience of participants in the management process. As youcarry out your professional activity, experience is formed, which subsequently allows you toperform some tasks much faster and more efficiently. To gain experience, it is proposed to usea knowledge management system based on ontology. In order to ensure maximum awarenessof the management of the situation, it is proposed to use the expert system of the situationcenter. (ES SC) The ES of the situational center of the authorities can significantly improvethe efficiency of management processes and provides information support for strategic andtactical management decision-making. The ES should implement tools for a comprehensiveand operational assessment of the state of the management object and situational analysis ofthe identified problems.
- 30 March 2022
- Published by Astana IT University in Scientific Journal of Astana IT University
- No. 9,p. 61-74
Open AccessINTELLECTUAL INFORMATION TECHNOLOGIES IN THE ACTIVITIES OF THE EMERGENCY RESCUE SERVICE
The paper considers the features of the use of intelligent information technologiesin the process of the emergency rescue service of the enterprise. To improve the efficiencyof the rescue services, it is necessary to carry out strategic planning and management,which would be designed to prevent the occurrence of an emergency (emergency). Effectivesubsystems of emergency proactive planning should not only predict the occurrence ofpossible emergencies, but also provide for appropriate preventive measures, and emphasisshould be placed on eliminating the underlying causes, not the emerging consequences. Thechange of modes in the event of an emergency can be effectively implemented through thedeployment of a technologically secured situational center. Decisions in emergency situationsare made in various operational situations, including crisis, and in extremely limited time.Nevertheless, they must be taken in a timely manner, be as reasonable as possible, and ensurethe fullest and most effective use of available opportunities. Decision-making processescan be based on the personal experience of participants in the management process. As youcarry out your professional activity, experience is formed, which subsequently allows you toperform some tasks much faster and more efficiently. To gain experience, it is proposed to usea knowledge management system based on ontology. In order to ensure maximum awarenessof the management of the situation, it is proposed to use the expert system of the situationcenter. (ES SC) The ES of the situational center of the authorities can significantly improvethe efficiency of management processes and provides information support for strategic andtactical management decision-making. The ES should implement tools for a comprehensiveand operational assessment of the state of the management object and situational analysis ofthe identified problems.