Traditional network traffic management systems are largely based on static rule that depend on predefined signatures by network security experts, which these systems perform well in detecting known traffic behaviours that matching with signatures rules, so they are ineffective with detect unknown anomalies traffic due to their rigid design. For addressing these challenges previous studies have utilized supervised learning methods such as Ran‎dom Forest an‎d Support Vector Machines (SVM) for traffic classification that achieving high accuracy in detecting known patterns. But they often fail when exposed to unseen traffic types. Although previous studies the best choice for detecting unseen pattern. But most existing works lack integration between real-time prediction an‎d deployment within a practical, user-friendly interface. Additionally, the challenge of generalizing models to han‎dle a wide variety of traffic types, practically those do not show during training, which remains a significant limitation. if, there is a need for a robust an‎d deployable solution that not only predicts traffic behaviour accurately but also adapts to improve network conditions in real time. For more securing network, our project appears the implementation of machine learning models for network traffic classification to enhance cybersecurity an‎d traffic monitoring. The specific objective of this project is to monitor an‎d classify network traffic into normal an‎d attack traffic. To do this, our project used different machine learning models such as Ran‎dom Forest. The result is that the ran‎dom forest classifier achieved the highest accuracy through demonstrating its effectiveness in han‎dling complex datasets with diverse features. The project involved preprocessing techniques that include scaling an‎d feature selec‎tion to optimize model performance. After that, building Flask interface with integrating a synthetic data generator an‎d a trained detection model for real-time classification. Some challenges are included aligning data dimensions an‎d computational requirements. Future work aims to incorporate deep learning models for enhanced prediction an‎d real-time network monitoring. This study confirms the potential of machine learning in improving network security an‎d operational efficiency.

10/29/2025 12:00:00 AM

87980

1404/08/09

ANOMALY DETECTION USING MACHINE LEARNING

Machine Learning , Anomaly Detection , Random Forest , Flask