According to the latest Verizon DBIR report, credential abuse, including password reuse and human factors in password creation, remains the leading attack vector. It was revealed that most users change their passwords only when they forget them, and 35% of respondents find mandatory password rotation policies inconvenient. These findings highlight the importance of combining technical solutions with user-focused education to strengthen password security. In this research, the “human factor in the creation of usernames and passwords” is considered a vulnerability, as identifying the patterns or rules used by users in password generation can significantly reduce the number of possible combinations that attackers need to try in order to gain access to personal data. The proposed method based on an LSTM model operates at a character level, detecting recurrent structures and generating generalized masks that reflect the most common components in password creation. Open datasets of 31,000 compromised passwords from real-world leaks were used to train the model and it achieved over 90% test accuracy without signs of overfitting. A new method of evaluating the individual password creation habits of users and automatically fetching context-rich keywords from a user’s public web and social media footprint via a keyword-extraction algorithm is developed, and this approach is incorporated into a web application that allows clients to locally fine-tune an LSTM model locally, run it through ONNX, and carry out all inference on-device, ensuring complete data confidentiality and adherence to privacy regulations.

Data breaches in healthcare are increasingly caused by insider threats, yet traditional logging systems remain vulnerable to tampering. This paper proposes a hybrid security framework for medical data in Kazakhstan, combining blockchain for immutable log storage with machine learning for automated anomaly detection. We introduce a dual-backend architecture where a private Ethereum network secures access logs, while a separate analysis module utilizes supervised learning algorithms to identify suspicious behaviour patterns in real-time. To validate the system, we generated a region-specific synthetic dataset compliant with Kazakhstani national identification standards (IIN). Experimental results demonstrate that the system effectively secures audit trails against modification and detects anomalous access patterns with high accuracy. The proposed solution addresses the critical gap between data integrity and proactive threat detection, offering a scalable architecture compliant with data privacy regulations.

Kazakh automatic speech recognition (ASR) presents a persistent challenge for large-scale multilingual models. This paper presents a systematic evaluation of 27 ASR pipeline configurations (three ASR models × three VAD methods × three post-processing strategies) on the Kazakh Speech Dataset (KSD), examining the contribution of model fine-tuning, voice activity detection (VAD) preprocessing, and large language model (LLM) post-correction and benchmarking the resulting pipelines against two non-Whisper foundation models. Language-specific fine-tuning reduces Word Error Rate (WER) from 43.20% (generic Whisper-large-v3) to 11.88% (Kazakh fine-tuned Whisper-turbo), a 31.32-percentage-point absolute reduction (72.5% relative; p < 0.001, bootstrap test); the effect persists after controlling for model size (generic Whisper-large-v3-turbo, 18.92%, vs. the same architecture after fine-tuning, 11.88%; p < 0.001). VAD preprocessing consistently degrades performance. Zero-shot post-correction with general-purpose LLMs yields no benefit and adds substantial latency: Gemma-2-9B and Qwen2.5-7B raise WER by 5.5 and 7.2 percentage points at real-time factors of 0.52 and 0.30, and a larger 32B model still degrades accuracy (+10.8 points), indicating that scale is not the limiting factor. Among all systems evaluated, a larger multilingual foundation model, SeamlessM4T-v2 (9.72% WER), outperforms the fine-tuned Whisper, showing that for Kazakh model coverage matters more than pipeline engineering. Character-level error analysis identifies systematic confusion between Kazakh-specific and Russian Cyrillic characters as a dominant error source. These findings establish that, for Kazakh under the evaluated conditions, model choice dominates pipeline add-ons: fine-tuning is essential, VAD and zero-shot LLM correction consistently hurt, and a strong multilingual model sets the best result; we further discuss the extent to which these conclusions extend to typologically similar Kipchak-Turkic languages.

The paper examines in detail the modern methods of image compression, focusing on how advanced algorithms are used in practical digital imaging systems. The study examines many compression methods, including LZMA, LERC, ZSTD and their mixed forms and compares how well they perform in terms of compression ratio, time required, memory efficiency and how much information entropy they keep. Machine learning methods for compression are used in the analysis, focusing on how they work with images from medical imaging as well as satellite data. Experiments are performed on standardized datasets, with the main goal of following the theoretical limits set by Shannon’s Source Coding Theorem. The study shows that using modern hybrid algorithms, it is possible to compress data by at least 4:1 and keep it safe, with LZMA and LERC combinations performing best when the data is subject to entropic constraints. The results show that using parallel processing leads to a 60% decrease in processing time when compared to traditional single-threaded methods. The results strengthen the theories and techniques needed for the next generation of compression systems, mainly for handling high-resolution images quickly.

Заманауи білім беру кеңістігінде жасанды интеллект элементтері бар білім беру платформалары оқытудың сапасын жекелендіру мен жақсартуда шешуші рөл атқарады. Бұл мақалада білім беру процесінде жасанды интеллект қолданатын білім беру платформасы архитектурасының тиімділігін бағалау әдістемесі талқыланады. Мұндай шешімдерді сәтті жүзеге асыруға әсер ететін негізгі параметрлер, соның ішінде жүйенің бейімделу қабілеті, пайдаланушылармен өзара әрекеттесу, мазмұнды жекелендіру деңгейі және аналитикалық мүмкіндіктер талданады. Зерттеуде талдау, жобалау, модельдеу және эксперименттік тестілеу әдістері қолданылып, нәтижелері шығарылды. Талдау әдісі қолданыстағы білім беру платформаларын зерттеу және олардың тиімділігіне әсер ететін негізгі факторларды анықтау үшін қолданылады. Жобалау әдісі жасанды интеллект интеграциясы бар білім беру платформасының архитектуралық моделін жасауға мүмкіндік береді. Модельдеу әртүрлі пайдалану сценарийлерінде жүйенің өнімділігін бағалау үшін пайдаланылады, ал эксперименталды тестілеу жиналған деректер негізінде оның тиімділігін эмпирикалық түрде тексеру үшін қолданылады. Алынған нәтижелер жасанды интеллектіні білім беру орталарында пайдаланудың маңыздылығын растайды және оқытушылар, білім алуышлар, білім беру технологияларын әзірлеушілер мен цифрлық оқытуды зерттеушілер үшін пайдалы болуы айдан анық. Кілтті сөздер: білім беру процесі, білім беру платформасы, оқыту, студент, жасанды интеллект.

Intelligent Aerospace Image Processing for Land use Identification and Smart Grid Integration