Real-Time Detection of Network Threats using Artificial Intelligence Techniques

Intrusion detection plays a vital role in safeguarding modern systems from evolving and complex cyber threats. Traditional approaches, such as Random Forest, Decision Trees, Auto encoders, and Generative Adversarial Networks (GANs), have shown limitations in adapting to dynamic attack patterns and ensuring high accuracy. To address these challenges, this study proposes a hybrid Intrusion Detection System (IDS) that combines GANs and Variational Auto encoders (VAEs). By leveraging the CICDDoS2019 dataset, the system incorporates effective preprocessing techniques like Synthetic Minority Oversampling Technique (SMOTE) to handle class imbalance and normalization to manage high-dimensional data. The GAN-VAE model excels at detecting anomalies by identifying deviations in network behavior, achieving a detection accuracy of 99%. This innovative, scalable, and real-time solution surpasses traditional methods, offering robust and adaptive protection against modern cyber-attacks.