Table 6 Performance comparison of CyberDetect-MLP with baseline models in IoT cyberattack detection.
Model/study | Approach | Dataset | Accuracy (%) | Scalability | Explainability |
|---|---|---|---|---|---|
Vinayakumar et al. (2019) | CNN + RNN | NSL-KDD, UNSW-NB15 | 97.01 | Moderate | Low |
Alrashdi et al. (2019) | ML (RF, SVM) | TON_IoT (subset) | 94.67 | Limited | Medium |
Ferrag et al. (2020) | Deep Learning Survey | Various IDS Datasets | 94–97 | Varies | Not applicable |
Shone et al. (2018) | Autoencoder + DNN | NSL-KDD | 96.21 | Low | Low |
Lopez-Martin et al. (2017) | CVAE (VAE) | Custom IoT dataset | 95.03 | Low | Medium |
Proposed CyberDetect-MLP | Optimized MLP | TON_IoT (full) | 98.87 | High | High (Grad-CAM) |
