Table 1 The performance of our deep learning models on the UNSW-NB15 dataset
From: Network intrusion detection using feature fusion with deep learning
DL model | Train | Validation Performance (%) | Test |
|---|---|---|---|
Early-fusion model | Â | Â | Â |
 Accuracy | 90.99 | 84.44 | 76.47 |
 Precision | 96.29 | 90.0 | 83.53 |
 Recall | 87.46 | 80.87 | 71.59 |
Late-fusion model | Â | Â | Â |
 Accuracy | 85.92 | 83.92 | 77.09 |
 Precision | 93.54 | 91.49 | 86.04 |
 Recall | 81.45 | 79.20 | 69.50 |
Late-Ensemble model | Â | Â | Â |
 Accuracy | 83.67 | 83.09 | 76.84 |
 Precision | 91.75 | 90.88 | 85.92 |
 Recall | 78.55 | 77.64 | 68.18 |