Enhancing Network Security through AI-Powered Anomaly Detection Using Generative Adversarial Networks

Developments in communication technology have facilitated more data sharing in geographically dispersed settings, but they have also enlarged the attack surface, raising questions about network security. Research focuses on AI-based anomaly detection systems to improve Network Intrusion Detection Systems (NIDSs) in order to address this. However, data imbalance makes it more difficult for AI models to learn and effectively identify threats when legitimate traffic outnumbers malicious traffic. To balance data classes and enhance NIDS performance, suggested options include employing Generative Adversarial Networks (GANs) to produce artificial data that mimics small attack traffic. In order to guarantee that AI models acquire enough training data for both legitimate and malicious network traffic, the method includes GANs. Methods such as “voting classifier” (RF+AB+DT) and “stacking classifier” (LightGBM+DT) are used to improve the system's accuracy and resilience in detecting threats. 

Keywords: Anomaly detection, Generative Adversarial Network (GAN), Network security, Network Intrusion Detection System (NIDS).