Table 7 Classification results for the proposed and the baseline models calculated with macro averaging
From: Model fusion of deep neural networks for anomaly detection
Method | Macro recall | Macro precision | Macro F1 score | Macro Fβ score | False alarm rate |
|---|---|---|---|---|---|
5-class DNN (baseline) (%) | 95 | 36.67 | 41 | 48.20 | 7.66 |
Model fusion (proposed) (%) | 90.33 | 48 | 55.33 | 65.18 | 2.30 |