Journal of Problems in Computer Science and Information Technologies

IMPROVING NETWORK INTRUSION DETECTION USING THE MINI-VGGNET ARCHITECTURE: TACKLING CHALLENGES OF IMBALANCED DATA.

Altynay Serdaly — 2024-12-14

In the field of cybersecurity, the detection of network intrusions is a pressing challenge, particularly when dealing with imbalanced datasets. This study presents a novel model based on the MINI-VGGNet architecture, tailored specifically for identifying various types of network attacks using the CICIDS2017 dataset. The objective is to enhance detection accuracy while effectively managing the challenges posed by imbalanced data. The proposed model incorporates convolutional layers to capture deep features from network data, allowing for improved classification of 15 distinct classes of attacks, including DoS and DDoS. Experimental results demonstrate that the model achieves high accuracy in classifying common attack types, although challenges remain in accurately identifying specific classes like Web Attack – XSS and SQL Injection. The architecture's efficiency and lower computational demands make it suitable for real-world applications, particularly in resource-constrained environments. The findings indicate that further refinement of data balancing techniques is necessary to improve classification performance across all attack types. Overall, this research showcases the effectiveness of the MINI-VGGNet-Intrusion model in advancing intrusion detection systems and highlights the ongoing need for innovation in methods for handling imbalanced cybersecurity datasets.

DEVELOPMENT OF INTELLIGENT WEB APPLICATIONS FOR AUTONOMOUS CONTROL OF ROBOTIC DEVICES (IN THE FORM OF MULTIDIMENSIONAL OBJECTS) IN THE AGRICULTURAL SECTOR

Aida Mustafayeva — 2024-12-14

Global agricultural challenges, including those faced in Azerbaijan, encompass climate change, water scarcity, irrigation difficulties, soil degradation, biodiversity loss, rising production costs, market uncertainty, labor shortages, and concerns around food safety and quality. These issues pose significant obstacles to the sustainable development of agriculture and demand innovative solutions. Climate change, for instance manifested – in rising temperatures and increased extreme weather events – creates an uncertain and often unfavorable environment for farmers. In Azerbaijan, the Presidential Decree dated July 15, 2021, on measures to promote agricultural production and processing, has established favorable conditions for the effective deployment of smart web applications designed for intelligent agricultural process control, as discussed in the article. The article explores several key areas of this decree, aligning them with the objectives of advancing smart technologies in agriculture. These proposed web-based solutions aim to enhance the automation, improvement, and efficiency of agricultural processes, and they are well-supported by government policies and incentives aimed at stimulating Azerbaijan’s agricultural sector.

ANALYSIS OF SPATIO-TEMPORAL CONVOLUTIONAL NEURAL NETWORKS FOR THE ACTION DETECTION TASKS

Nurtugan Azatbekuly — 2024-12-14

This study investigates the effectiveness of Spatio-Temporal Convolutional Neural Networks (ST-CNNs) for action detection tasks, with a comprehensive comparison of state-of-the-art models including You Only Watch Once (YOWO), YOWOv2, YOWO-Frame, and YOWO-Plus. Through extensive experiments conducted on benchmark datasets such as UCF-101, HMDB-51, and AVA, we evaluate these architectures using metrics like frame-based Mean Average Precision (frame-mAP), video-mAP, computational efficiency (FPS), and scalability. The experiments also include real-time testing of the YOWO family using an IP camera and RTSP protocol to assess their practical applicability. Results highlight the superior accuracy of YOWO-Plus in capturing complex spatio-temporal dynamics, albeit at the cost of processing speed, and the efficiency of YOWO-Frame for live applications. This analysis underscores the trade-offs between speed and accuracy inherent in single-stage ST-CNN architectures. Our findings from the comparative analysis provide a robust foundation for the development of real-time systems capable of efficient and reliable operation in action detection tasks.

ANALYSIS OF THE EFFECTIVENESS OF OBJECT RECOGNITION METHODS IN IMAGES

Zhansaya Segizbayeva — 2024-12-14

This paper considers the problem of object recognition in images, which is one of the key problems of computer vision. The relevance of the research is due to the wide application of object recognition systems in such areas as security, medicine, robotics, automotive industry and quality control. The research analyses existing recognition methods, including traditional approaches and modern deep learning methods. Their advantages, disadvantages and effectiveness in different environments are evaluated. On the basis of experimental data, the most effective algorithms for application in recognition systems were selected. The results of the work allowed us to propose recommendations for the selection and improvement of methods of object recognition, which helps to improve the accuracy and reliability of such systems. The obtained conclusions can be useful for specialists in the field of computer vision and developers of applications that use recognition technologies.

USING SYNTHETIC DATA TO IMPROVE DATA PROCESSING ALGORITHMS IN BUSINESS INTELLIGENCE

Aizat Dildabek — 2024-12-14

The growing volumes of data require the development of effective methods for its processing to solve practical problems. This study is devoted to the use of synthetic data to improve data processing algorithms in business analysis tasks. Synthetic data has a number of benefits, including increasing the amount of data available to train models and ensuring privacy when working with sensitive financial and medical data. The paper examines the potential of synthetic data generated by CTGAN and TVAE methods for regression problems. The study uses two datasets—Health Insurance and Boston Housing—to evaluate the performance of machine learning models, such as linear regression, random forest, and gradient boosting. The results suggest that synthetic data can significantly improve algorithm performance, especially for small or unbalanced datasets, although challenges remain in achieving quality comparable to real-world data. The study highlights the practical importance of synthetic data for optimizing business processes and opens up new opportunities for further study of data generation methods and their application.