Distributed Denial-of-Service (DDoS) attacks pose a significant threat to online infrastructure, with evolving tactics that often evade conventional detection methods. The proposed system employs a 3D convolutional neural network (3D CNN) combined with hive-plot sequences to effectively classify DDoS traffic, achieving over 93% accuracy on benchmark datasets. This method not only establishes a robust baseline for pattern recognition but also incorporates adversarial training to further enhance its resilience against manipulation.
The research highlights the importance of early-stage classification by analyzing frame-wise predictions, which can provide crucial signals for timely detection of attacks. The implications of this advancement are considerable for cybersecurity measures, as maintaining high accuracy in both adversarial and clean samples is essential in real-world applications. As DDoS methods continue to adapt, such innovative approaches will be critical in defending against increasingly sophisticated cyber threats.
👉 Pročitaj original: arXiv AI Papers
