Computer Science, Computer Engineering, Artificial Intelligence, Information Systems
9
Scopus Publications
Scopus Publications
Integrated IoT–UAV Architecture for Three-Dimensional Electromagnetic Radiation Monitoring and Intelligent Source Classification Saken Mambetov, Dinara Nurpeissova, Kyrmyzy Taissariyeva, Gulnara Tleuberdiyeva, Zhanna Mukanova, Bakhytzhan Kulambayev, Altynbek Moshkalov, Aigul Skakova Electronics Switzerland, 2026 The rapid deployment of 5G networks and the proliferation of Internet of Things (IoT) devices have significantly increased the complexity of urban electromagnetic radiation (EMR) environments. Conventional ground-based monitoring systems are spatially limited and unable to provide three-dimensional field characterization. This paper proposes an integrated IoT–UAV framework for high-resolution EMR monitoring, spatial reconstruction, and intelligent source classification. A four-layer architecture combining distributed sensing, edge computing, cloud analytics, and visualization is developed. A formal electromagnetic propagation model is introduced to ensure consistency between broadband exposure measurements and frequency-selective spectral analysis. A CNN–LSTM architecture is implemented for spectral–temporal source classification, achieving 95% validation accuracy across five EMR categories. Simulation-based validation demonstrates up to an eightfold improvement in spatial coverage compared to fixed ground networks while maintaining a practical anomaly detection threshold of −55 dBm in the spectrum-analysis RF chain. The proposed framework establishes a mathematically consistent and practically deployable solution for next-generation EMR monitoring systems.
Methods for automatic emotion recognition in hacker forum texts Ceur Workshop Proceedings, 2026
Design and Empirical Evaluation of a Four-Layer AI Agent Architecture for Automated Web Application Security Testing Bakhytzhan Kulambayev, Gulnar Astaubayeva, Zhanna Mukanova, Kuralay Makhmetova, Saken Mambetov, Serik Joldasbayev Engineering Technology and Applied Science Research, 2026 This study proposes a four-layer AI agent architecture for automating routine web security operations, integrating Large Language Model (LLM) reasoning with a hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) detection engine and implementing a Reasoning-Acting (ReAct) loop for autonomous testing with human-in-the-loop validation. The proposed architecture was empirically evaluated across 50 web applications sourced from OWASP WebGoat, DVWA, and custom-developed test environments over a six-month period. The experimental results demonstrate that the AI agent achieved an overall detection accuracy of 89.2% (95% CI: 86.4-92.0%), significantly outperforming traditional automated methods (67.4% accuracy, p < 0.001). Mean Time to Remediation (MTTR) decreased from 74.3 days to 28.5 days (61.6% reduction), while false positive rates decreased from 24.3% to 4.8%. According to these findings, AI agent-driven automation can substantially enhance the efficiency and reliability of web security testing. However, human expertise remains important for assessing complex vulnerabilities and detecting zero-day threats.
Graph-based neural networks with neural ODEs for robust speech processing in environmental and human-centric systems Saken Mambetov, Ainur Ormanbekova, Serik Joldasbayev, Laura Duissembayeva, Bulgyn Mailykhanova E3s Web of Conferences, 2025 This paper introduces H-STGNN-ODE-DA, a novel model for voice sentiment analysis that combines multi-scale acoustic feature extraction, hierarchical graph neural networks (GNNs), Neural Ordinary Differential Equations (Neural ODEs), and domain-adversarial adaptation. Designed to enhance accuracy and robustness under real-world conditions, the model was evaluated on IEMOCAP, MELD, and EmoDB datasets, outperforming state-of-the-art approaches such as LSTM, CNN, GCN, GAT, DANN, and SPECTRA. Notably, it achieved a 4.0% improvement over GAT on MELD. Neural ODEs enabled effective modeling of continuous emotional transitions, while domain-adversarial adaptation ensured robustness to domain shifts. Ablation studies confirmed the critical role of each component in achieving high performance. The model demonstrated strong cross-domain transferability, maintaining high accuracy across diverse recording conditions. These results position H- STGNN-ODE-DA as a robust and versatile solution for real-world applications in speech processing, including virtual assistants, social media analysis, and customer service systems.
Development of a Framework for Analyzing Kazakh Language Errors in Semi-Structured Data Gaukhar Munaitbas, Laura Baitenova, Saken Mambetov, Saule Tussupova Procedia Computer Science, 2025 This paper presents a novel hybrid framework for systematic error analysis in semi-structured Kazakh texts, addressing the dual challenges of rich agglutinative morphology and low-resource data environments. The proposed pipeline integrates deterministic linguistic rules derived from the Apertium Turkic lexicon with probabilistic classifiers trained on an iteratively expanded corpus of 120 198 sentences. Core components include text normalization, tokenization, lexicon lookup, morphological parsing, and error classification, each formalized through mathematical functions to ensure reproducibility. Empirical evaluation against baseline systems demonstrated an F₁-score improvement of 12 percentage points, with preprocessing accuracy rising from 89 % to 96 % following successive lexicon augmentations. Cross-platform analysis of 3 000 records revealed that Russian-spelling interference accounts for nearly half of all detected errors, whereas typographical and pure spelling mistakes vary substantially across user-generated and editorial sources. Comprehensive linguistic resources—comprising over 70 000 root forms, 5 136 affix entries, and a validated correct-word dictionary—were compiled to underpin robust error detection. While manual lexicon enrichment proved effective, its labor intensity suggests the need for semi-supervised morphological induction in future work. Overall, this study establishes a scalable, adaptable approach for error analysis in agglutinative languages and lays the groundwork for extending these methods to other Turkic language contexts.
Hybrid artificial intelligence architectures for automatic text correction in the Kazakh language Laura Baitenova, Saule Tussupova, Saken Mambetov, Gauhar Munaitbas, Gulnar Mukhamejanova Frontiers in Artificial Intelligence, 2025 The Kazakh language, as an agglutinative and morphologically rich language, presents significant challenges for the development of natural language processing (NLP) tools. Traditional rule-based analyzers provide full coverage but lack flexibility; statistical and neural models handle disambiguation more effectively, yet require large annotated corpora and substantial computational resources. This paper presents a hybrid morphological analyzer that integrates Finite-State Transducers (FST), Conditional Random Fields (CRF), and transformer-based architectures (KazRoBERTa, mBERT). For the experiments, a new corpus, KazMorphCorpus-2025, was created, consisting of 150,000 sentences from diverse domains annotated for morphological analysis. Experimental evaluation demonstrated that the KazRoBERTa model consistently outperforms mBERT in terms of accuracy, F1-score, and prediction speed. The hybrid architecture effectively combines the exhaustive coverage of FST with the contextual disambiguation of neural networks, reducing errors associated with homonymy, borrowings, and long affixal chains. The results confirm that the proposed system achieves a balance between accuracy, efficiency, and scalability. The study underscores the practical significance of hybrid approaches for tasks such as spell checking, information retrieval, and machine translation in the Kazakh language, as well as their potential transferability to other low-resource Turkic languages. Future work will include the expansion of the corpus, integration of KazBERT and mBERT models, and validation of the proposed approach in applied NLP systems.
DETECTION AND CLASSIFICATION OF THREATS AND VULNERABILITIES ON HACKER FORUMS BASED ON MACHINE LEARNING Saken Mambetov, Ihor Ilhe, Vitalina Babenko, Bakytzhan Kulambayev, Olena Fridman, Serik Joldasbayev, Hanna Doroshenko, Oleksandr Gurko, Yenlik Begimbayeva, Serhii Neronov Eastern European Journal of Enterprise Technologies, 2024 The object of this study is the process of detecting threats and vulnerabilities in hacker forums, which are a well-known source of potential dangers for Internet users. However, the problem of analyzing and classifying data from these forums is its complexity due to such features of the participants' language as specific slang, jargon, etc., which requires the use of modern tools of their processing. This paper explores the application of machine learning to devise an effective method for analyzing sentiment and trends in hacker forums to identify potential threats and vulnerabilities in cyberspace. All necessary stages of the process of detecting threats and vulnerabilities have been developed, ranging from data collection and preprocessing to the training of a model that is capable of processing “raw” unstructured data from hacker forums. The implementation of six popular machine learning algorithms, namely k Nearest Neighbors (kNN), Random Forest, Naive Bayes, Logistic Regression, Support Vector Machines (SVM), and Decision Tree algorithms have been studied with a view to determining their efficiency of threat and vulnerability detection and classification. The experiments have been conducted on real data (150,000 messengers). It has been determined that the Random Forest algorithm coped with the task the best (accuracy=0.89, recall=0.84, precision=0.91, F1-score=0.87 and ROC-AUC=0.89). The proposed tool based on machine learning not only collects data that poses a potential threat but also processes and classifies it according to the specified keywords. This allows detecting threats and vulnerabilities at a high speed. The results of the study make it possible to identify potential trends in threats and vulnerabilities. This will contribute to the improvement of cybersecurity systems and ensure more reliable protection of information resources
Internet threats and ways to protect against them: A brief review Saken Mambetov, Yenlik Begimbayeva, Serik Joldasbayev, Gulnur Kazbekova Proceedings of the 13th International Conference on Cloud Computing Data Science and Engineering Confluence 2023, 2023 Since the 21st century is the century of information technology, there are now a large number of Internet users, and the number of these users is steadily growing. Most people know the useful aspects of this network, but there are dangerous aspects of the Internet, which are reviewed in this paper. Today there are many types of threats such as phishing attacks, spam messages, malware, worms, spyware, Trojans, rooters, botnets. The main principle of protection against these threats is information and cyber literacy of using the Internet. The main purpose of this article is a summary of Internet threats based on a review of articles by previous researchers, as well as the definition of types of threats and attacks. As a result of the study, threats were filtered into two main types: threats of social engineering, where the main emphasis is on the information carrier, and technical threats, where various methods, algorithms and software implementations are used to hack directly into a computing device containing information.
Solving the Problem of Discrete Process Control Synthesis Using Optimization on a Sliding Interval Zhazira Julayeva, Waldemar Wójcik, Gulzhan Kashaganova, Күлжан Тогжанова, Saken Mambetov International Journal of Electronics and Telecommunications, 2023 — The paper presents a solution to the problem of synthesis of control with respect to the sliding interval length for the optimization of a class of discrete linear multidimensional objects with a quadratic performance criterion. The equation of motion of a closed multidimensional discrete system in the general non-stationary case is derived based on the length of the optimization interval and their main properties. The closed-loop is fitted with a signal representing the predicted values averaged over the whole sliding interval of optimization with a certain weight. A problem with a sliding optimization interval may not require a real-time solution by means of a sequence of solutions on compressed intervals. Therefore, the study of control systems with optimization on a sliding interval is of undoubted interest for a number of practically important control problems.
