Intrusion Detection System Based on ViTCycleGAN and Rules

This paper explores the application of deep learning techniques in the field of intrusion detection and its potential problems. Possible challenges of deep learning in intrusion detection include handling the imbalance between positive and negative samples, which leads to unstable model performance in distinguishing normal and abnormal traffic. To address this issue, the paper proposes combining deep learning-based intrusion detection techniques with a rule-based approach to enhance the system's adaptability and intelligence. The specific scheme includes four Vision Transformer models, two generators, and two discriminators. The discriminators are used to differentiate normal traffic and detect abnormal behaviors, following strict detection rules to reach a final determination. Through validation on the NSL-KDD dataset and CIC-DDOS2019 dataset, the proposed scheme achieves accuracies of 98.32% and 99.23%, respectively, providing new research insights and solutions in the field of intrusion detection.