Abstract

Machine Learning-Based Anomaly Detection for Secure Enterprise Network Infrastructure focuses on applying machine learning techniques to identify unusual network behavior and potential security threats within complex enterprise environments. Modern organizations generate large volumes of network traffic, logs, authentication records, and system events, making conventional signature-based security mechanisms insufficient for detecting previously unknown or rapidly evolving attacks. Machine learning-based anomaly detection addresses this challenge by learning patterns of normal network behavior and identifying significant deviations that may indicate malicious activity, system misuse, configuration errors, or operational failures. This study examines the application of supervised, unsupervised, and semi-supervised learning techniques for enterprise network security, including clustering, classification, isolation-based methods, and deep learning approaches. The proposed framework can integrate data collected from network devices, endpoints, servers, firewalls, intrusion detection systems, and security logs to construct a comprehensive behavioral profile of enterprise infrastructure. Feature engineering and dimensionality-reduction techniques can be used to improve detection efficiency, while adaptive learning mechanisms can help systems respond to changing traffic patterns and emerging threats