Fire detection is an important task in the field of safety and emergency prevention. In recent years, deep learning methods have shown high efficiency in solving various computer vision problems, including detecting objects in images. In this paper, monitoring wildfires was considered, which allows you to quickly respond to them and prevent their spread using deep learning methods. For the experiment, images from the satellite and images from the FireWatch sensor were taken as initial data. In this work, the deep learning algorithms you only look once (YOLO), convolutional neural network (CNN), and fast recurrent neural network (FastRNN) were considered, which makes it possible to determine the accuracy of a natural fire. As a result of the experiments, an automated fire recognition algorithm using YOLOv4 deep learning methods was created. It is expected that the results of the study will show that deep learning methods can be successfully applied to detect fire in images. This may lead to the development of automated monitoring systems capable of quickly and reliably detecting fire situations, which will help improve safety and reduce the risk of fires.

This work is devoted to the research and development of methods for effectively identifying breast pathologies using modern machine learning technologies, such as you only look once (YOLOv8) and faster region-based convolutional neural network (R-CNN). The paper presents an analysis of existing approaches to the diagnosis of breast diseases and an assessment of their effectiveness. YOLOv8 and Faster R-CNN architectures are then applied to create pathology detection models in mammography images. The work analyzed and classified identified breast pathologies at six levels, taking into account different degrees of severity and characteristics of the diseases. This approach allows for more accurate determination of disease progression and provides additional data for more individualized treatment planning. Classification results at various levels can improve the quality of medical decisions and provide more accurate information to doctors, which in turn improves the overall efficiency of diagnosis and treatment of breast diseases. Experimental results demonstrate high accuracy and speed of image processing, providing fast and reliable detection of potential breast pathologies. The data obtained confirm the effectiveness of the use of machine learning algorithms in the field of medical diagnostics, providing prospects for the further development of automated systems for detecting breast diseases in order to improve early diagnosis and treatment efficiency.

The classification of the natural and anthropogenic destabilizing factors of a telecommunications network as a complex system is presented herein. This research shows that to evaluate the parameters of a telecommunications network in the presence of destabilizing factors, it is necessary to modify classical linear methods to reduce their sensitivity to the incompleteness of a priori information. Using generalized linear models of multiple regression, a combined method was developed for assessing and predicting the survivability of a telecommunications network under conditions of uncertainty regarding the influence of destabilizing factors. The method consists of accumulating current information about the parameters and state of the network, the statistical analysis and processing of information, and the extraction of sufficient sample statistics. The basis of the developed method was balancing multiple correlation–regression analysis with the number of regression equations and the observed results. Various methods of estimating the mathematical expectation and correlation matrix of the observed results under the conditions of random loss of part of the observed data (for example, removing incomplete sample elements, substituting the average, pairwise crossing out, and substituting the regression) were analyzed. It was established that a shift in the obtained estimates takes place under the conditions of a priori uncertainty of the statistics of the observed data. Given these circumstances, recommendations are given for the correct removal of sample elements and variables with missing values. It is shown that with significant unsteadiness of the parameters and state of the network under study and a noticeable imbalance in the number of regression equations and observed results, it is advisable to use stepwise regression methods.

This paper discusses an adaptive method of image steganography issues based on the application of a linear hash function over the GF (2) field to control the embedding process. The method uses staggered splitting of an image into 8 × 8-pixel blocks to provide blind steganography. Classification thresholds are defined as the percentiles of the distribution of gradients throughout the image, allowing for efficient load distribution between textured and smooth areas. Experiments on the BOSSBase, SIPI and Kaggle kits show that the method provides an actual capacity of up to 0.7 bpp at PSNR 47–50 dB and is resistant to statistical tests and RS analysis. At the same time, like other approaches based on modification of pixel differences, it remains vulnerable to modern stegoanalysis based on spatial rich models (SRMs). However, thanks to the modular structure of embedding control based on linear hash function, the proposed architecture allows direct integration with many modern adaptive strategies aimed at minimizing statistical anomalies.

This paper examines the application of machine learning methods for automatic text processing of abstracts from scientific articles. With the increasing volume of scientific information, researchers are faced with the problem of information overload, which makes it difficult to find and analyze relevant materials. To solve this problem, we are implementing machine learning algorithms such as the Support vector Machine (SVM) method and word representation using Word2Vec, which allows us to effectively classify annotations and extract key information. In the process, we collect data from open databases. Annotations go through preprocessing stages, including tokenization, lemmatization, and deletion of stop words. Then we use Word2Vec to convert annotation texts into vector representations, which serve as input data for the SVM model. The effectiveness of the models is evaluated using accuracy, completeness and F1-measure metrics. The results show that the integration of SVM and Word2Vec significantly improves the quality of annotation classification, which makes it possible to speed up the process of searching for scientific information. The work highlights the potential of using machine learning methods to automate the processing of scientific texts and suggests areas for further research, including the use of more complex models such as transformers. This methodology can become the basis for the development of effective tools that facilitate faster knowledge sharing in the scientific community.

In recent decades, the use of virtual reality (VR) and augmented reality (AR) in the field of education has been growing significantly. Virtual reality (VR) and augmented reality (AR) are becoming one of the key directions in the digital transformation of education in Kazakhstan. Their immersive and interactive nature offers teachers and students unique opportunities to collaboratively explore complex concepts and environments. These technologies open new possibilities for improving the quality of the learning process, expanding access to modern educational resources, and developing practical skills that are in demand in a rapidly changing economy. Using VR, it is possible to create fully immersive learning environments in which students can safely simulate real-life scenarios-from conducting laboratory experiments to mastering complex technical processes. Such learning helps develop critical thinking, increases motivation, and enhances understanding of the material due to the strong sense of presence. AR technologies, in turn, make the learning process more visual and interactive by overlaying digital information onto the surrounding physical world. In schools and universities across Kazakhstan, the use of AR is already showing high effectiveness: students can visualize anatomical structures, geological layers, historical artifacts, and other objects that are difficult to understand. This makes abstract concepts easier to comprehend and allows students to maintain their individual learning pace. In recent years, Kazakhstan has been actively implementing state and private initiatives aimed at integrating VR/AR technologies into the educational process. Specialized laboratories, digital classrooms, and pilot projects in engineering, medical, and humanities programs are being launched. The development of these technologies helps build an innovative educational ecosystem, enhance digital literacy, and develop the competencies required for the future among young people. To achieve this goal, the article analyzes these initiatives and explores how digital tools contribute not only to subject-specific knowledge but also to the development of transversal competencies. Among the selected examples, this article highlights the pedagogical value of virtual and augmented reality technologies, their advantages, and their practical application in the university environment. The findings demonstrate that immersive technologies make the learning process engaging, flexible, and student–centered. Thus, VR and AR are becoming powerful tools for modernizing Kazakhstan’s education system. Their integration helps create a flexible, accessible, and practice-oriented learning model that aligns with the requirements of the digital economy and the country's strategic development priorities.