Hybrid Intrusion Detection System

The purpose of (IDS) is to protect digital data from cyber threats. By quickly distinguishing and reacting to risks, IDS minimizes possible damage to systems and data. However, it can suffer from performance issues, such as delayed warnings and compromised accuracy. To solve this problem, our approach combines IDS with machine learning and a deep learning approach. It uses agglomerative hierarchical clustering and random forest (RF) algorithms for classification, supported by recurrent neural networks (RNN) for attack classification. Their implementation in MATLAB allows the classification of events. Evaluation with the NSL-KDD and CIS-IDS2017 datasets showed a high true positive rate (TPR) and successful detection of minority attacks, confirming its effectiveness.