The purpose of the study is to solve an extreme mathematical problem—semantic analysis of natural language, which can be used in various fields, including marketing research, online translators, and search engines. When training the neural network, data training methods based on the latent Dirichlet allocation model and vector representation of words were used. This study presents the development of a neurocomputer system used for the purpose of semantic analysis of the text in the Kazakh language, based on machine learning and the use of the latent Dirichlet allocation model. In the course of the study, the stages of system development were considered, regarding the text recognition algorithm. The Python programming language was used as a tool using libraries that greatly simplify the process of creating neural networks, including the Keras library. An experiment was conducted with the involvement of experts to test the effectiveness of the system, the results of which confirmed the reliability of the data provided by the system. The papers of modern computer linguists dealing with the problems of natural language processing using various technologies and methods are considered.

The object of the study is a neural network model for forecasting the dynamics of the number of grain pests. The main attention is paid to the verification of the artificial neural network model, on the basis of which the analytical system was developed. The paper solves the problem of developing a comprehensive method for verifying artificial neural network models. The proposed method combines the analysis of invariant sets, probabilistic assessment of stability, scenario testing and statistical assessment of the accuracy of the model. Unlike traditional methods based on calculations of the mean square error, the developed approach allows for a formal assessment of the probability of forecasts going beyond acceptable limits and analyzes the influence of external factors, such as temperature and humidity, on the accuracy of the forecast. Experimental validation was carried out using data on grain pest populations in the Republic of Kazakhstan from 2005 to 2022. The transformer-based prognostic model was trained on climatic and agronomic data. Model verification showed that the probability that the predictions will remain within the invariant set was 95.2%, with MSE 239.1234, determination coefficient R2 = 0.85. Scenario analysis showed that the model remained stable in the temperature range [15°C, 30°C], but the accuracy decreased under extreme conditions. The results indicate that the proposed complex method provides a more accurate assessment of the stability of forecasting compared to traditional approaches. The results can serve as a basis for further development of similar forecasting systems that can be used in agromonitoring and planning of crop protection measures

Forecasting crop pest outbreaks under conditions of increasing agroclimatic variability is a critical task for intelligent decision support systems in agriculture. Traditional statistical and empirical models typically have limited transferability and insufficient accuracy when describing nonlinear and multiscale relationships between climatic factors and pest population dynamics. This paper proposes a hybrid algorithm combining wavelet analysis and deep learning methods for forecasting agroclimatic pest infestation levels. The algorithm is based on multiscale decomposition of time series using a discrete wavelet transform, after which the extracted components are used as input features for a deep neural network implementing a nonlinear mapping between climatic parameters and infestation indicators. The developed computational framework includes the stages of data preprocessing, feature space formation, model training, and forecast generation in a single, reproducible pipeline. An experimental evaluation using long-term agroclimatic and phytosanitary data showed that the proposed algorithm outperforms classical regression and individual neural network models in terms of RMSE, MAE, and the coefficient of determination. The results confirm the effectiveness of integrating wavelet analysis and deep learning for developing phytosanitary risk forecasting algorithms and demonstrate the potential of the proposed approach for implementation in intelligent precision farming systems.

The object of this study is an autonomous Raspberry Pi-based device for real-time pest detection. The task addressed relates to the lack of affordable, energy-efficient, and autonomous solutions for working in the field without an Internet connection. The paper reports the design of an intelligent device for pest monitoring. The device is focused on automatic recognition of the striped grain flea beetle (Phyllotreta vittula) in grain crops. As a result of the study, a system was designed based on the Raspberry Pi 4.0 microcomputer using the OpenCV library and the YOLO model. The device processes the video stream, identifies pests, and saves data locally. The system provides high accuracy at low power consumption. This was made possible by a lightweight neural network architecture and optimized image processing. A distinctive feature of the solution is autonomy, mobility, and resistance to variable lighting conditions. The system also works with limited computing resources. The results demonstrate that the device could be effectively used in precision farming systems and at scientific institutions. The device helps identify pests and make agricultural decisions at early stages of infection. The technological advancement could be adapted to other types of pests with minimal changes to the model. In the future, the system could be integrated into broader agricultural monitoring platforms with the ability to transfer data to the cloud. The practical use of the device is possible both in large farms and on private farms. This technological advancement is especially relevant for regions with limited technical infrastructure

Structural health monitoring systems generate large volumes of multichannel dynamic data under external impacts, which require efficient processing methods for timely damage detection. This study aimed to develop and evaluate a parallel approach for processing structural monitoring data and diagnosing structural conditions using machine learning techniques. The proposed methodology included signal preprocessing, segmentation, normalization, and extraction of statistical, energy, and spectral features from multichannel vibration and deformation signals. A Random Forest classifier was trained to distinguish between the undamaged and damaged structural states. The model was implemented in Python using signal processing and machine learning libraries, while parallel data processing was applied at the levels of channel analysis, time-window processing, and feature extraction to reduce computational complexity. The experimental results demonstrated high classification performance. The developed model achieved a hold- out ROC–AUC of 1.000, indicating strong separability between the structural states. Five-fold cross-validation further confirmed the robustness of the proposed approach. Feature importance analysis showed that spectral characteristics, particularly bandwidth-related features, made the greatest contribution to classification accuracy. The obtained results also demonstrated that parallel processing significantly improved computational efficiency when analyzing large multichannel datasets. The study confirmed that the integration of signal processing, machine learning, and parallel computing provided an effective framework for structural condition diagnostics under dynamic loading. The proposed approach can be applied in structural health monitoring systems for early damage detection and near real-time analysis of monitoring data. Future research will focus on expanding the dataset and improving the generalization capability of the model.