مشاريع طلاب ربيع 2024 - S24
اكتشاف الشذوذ في مجموعة البيانات الشبكية باستخدام تقنيات التعلم الآلي
Anomaly Detection In Network Dataset Using Machine Learning Techniques
Anomaly detection in network data using machine learning is a powerful technique for identifying unusual patterns or behaviors in network traffic that deviate from expected norms. This approach leverages various machine learning methodologies to analyze data streams, enabling early warning systems for critical infrastructure and enhancing network security. Unsupervised methods such as clustering and density-based algorithms can identify anomalies without labeled data, while supervised methods involving classification algorithms, ensemble models, and deep learning can be trained on labeled data to classify traffic as normal or anomalous. However, challenges remain in discovering the underlying features that affect the accuracy of anomaly detection and the time required to detect anomalies. Despite these challenges, anomaly detection using machine learning remains a core component of modern network security and management strategies, and is constantly evolving to address emerging threats and complex network environments. Therefore, in this research, the effect of features on machine learning algorithms was studied using three feature selection algorithms: Feature Importance, correlation, chi-square, and three machine learning classifiers: decision tree, random forest, and nearest neighbor algorithm. The results showed that the features selected using the FI algorithm are the best of the rest of the algorithms, as the random forest and decision tree classifier were the best of the rest of the classifiers with an f1 score of 1.
إعداد: الطالب حسين ابراهيم علي
إشراف: الدكتور أكرم مرعي
اكتشاف الشذوذ في مجموعة البيانات الشبكية باستخدام تقنيات التعلم الآلي