MACHINE LEARNING APPROACHES FOR DETECTING REAL-TIME NETWORK IRREGULARITIES IN ENTERPRISE SYSTEMS
Mirzayev M. M. & Egamberdiyev M. M.
Annotatsiya
This study presents the design and assessment of a machine learning–driven approach for identifying unusual network behavior in enterprise environments. The framework applies both supervised and unsupervised learning techniques—such as Random Forest classification, K-Means clustering, and an Autoencoder-based model—to differentiate malicious activity from routine traffic. The objective is to detect security breaches, unauthorized access, malware activity, and other forms of network disruption. The evaluation was carried out using the NSL-KDD and CICIDS datasets, allowing systematic comparison of detection accuracy, scalability, false-alert frequency, and performance on encrypted communication streams. Recent evaluations show that using an Autoencoder to extract traffic characteristics, combined with Random Forest–based decision logic, achieves more reliable results than standard monitoring approaches. This combined method improves accuracy and noticeably lowers the number of false notifications.Taken together, the findings highlight that machine learning can serve as a strong and adaptable framework for protecting large organizational networks, supporting early threat recognition and ongoing security improvement.
Adabiyotlar
- Goni, Osman. (2021). Design, Deployment and Implementation of Local Area Network (LAN) at BAEC Head Quarter. International Journal of Computer Science and Mobile Computing. 10. 15-23. 10.47760/ijcsmc. 2021.v10i05.003.
- Djalilov, M., Burxonova, M., Suyumov, J., Polvonov, A., & Abduvaliyev, I. (2024). The compelled fluctuations of the rectangular two-layer a piecewise homogeneous plate of the constant thickness. In E3S Web of Conferences (Vol. 508, p. 04009). EDP Sciences.
- Mirzayev, M. (2025). Ensuring network security using cisco technology. Research and Implementation, 3(10), 75–79.
- Mirzayev, M. (2025). A comprehensive security protocol architecture for fully digitalized renewable energy markets. Research and Implementation, 3(12), 185–189.
- Mirzayev, M. (2025). A comprehensive analysis of acl standards and standard access control lists for network traffic filtering. Research and Implementation, 3(10), 567–571.