ARCHIVES

The exponential growth of digital networks has increased the risk of cyber-attacks, making network anomaly detection a critical component of modern cybersecurity infrastructure. Traditional Intrusion Detection Systems (IDS) rely heavily on rule-based or signature-based mechanisms, which are limited in their ability to identify novel or evolving threats and often generate high false-positive rates. This study, Network Anomaly Detection, proposes a machine learning–based framework to detect and classify abnormal traffic patterns with high accuracy and minimal false alarms. The NSL-KDD dataset, a widely used benchmark for intrusion detection, was employed for model training and evaluation. Data preprocessing techniques, including label encoding, normalization, and feature selection, were applied to improve model efficiency. Multiple supervised learning algorithms, such as Random Forest, Logistic Regression, and ensemble models, were implemented and compared. Performance was assessed using metrics such as accuracy, F1-score, confusion matrix, and ROC-AUC. A real-time web application was developed using Streamlit to provide end-users with an interactive interface for anomaly detection. The results demonstrate that the proposed framework offers a scalable, accurate, and user-friendly solution for identifying cyber threats, highlighting the role of machine learning in advancing beyond the limitations of traditional IDS approaches