A generative adversarial network-augmented dual-stream Mamba-transformer framework for advanced intrusion detection and cyber threat mitigation in smart healthcare networks

These limitations highlight the need for a unified, efficient, and robust intrusion detection framework capable of addressing heterogeneity, temporal modeling, imbalance, and computational constraints simultaneously.

The proposed framework is evaluated on the CICIoMT2024 dataset and compared with several baseline models, including MLP, CNN, BiLSTM, Transformer-only, and Mamba-only architectures.

Ben et al.13 introduced IDS-GraphMamba, an innovative intrusion detection framework that combines Mamba-style selective state-space modeling with Markov-based aggregation.

The arp_spoofing class (class 0) achieves a recall of 78.91% and an F1-score of 0.8059, reflecting the benefit of CTGAN augmentation for this low-frequency spoofing category.

In addition, attack labels were harmonized by mapping dataset-specific attack categories to a consistent multi-class intrusion detection framework while preserving normal traffic labels.