Table 6 IDS performance on CIC-IoT2023 (multi-class classification of 8 classes: 7 attacks + normal).
From: Dataset-centric evaluation of federated intrusion detection models in IoT networks
FL algorithm | Model | Accuracy | Precision | Recall | F1-score | AUC |
|---|---|---|---|---|---|---|
FedAvg | LSTM | 95.0 | 93.0 | 91.0 | 92.0 | 96.0 |
FedAvg | Transformer | 96.0 | 94.0 | 92.5 | 93.2 | 97.0 |
FedProx | LSTM | 95.5 | 93.5 | 92.0 | 92.7 | 96.5 |
FedProx | Transformer | 96.5 | 95.0 | 93.5 | 94.2 | 97.5 |
FedNova | LSTM | 96.0 | 94.0 | 93.0 | 93.5 | 97.0 |
FedNova | Transformer | 97.0 | 96.0 | 94.5 | 95.2 | 98.0 |
