Research Article
Effective Anomaly Detection Using Deep Learning in IoT Systems
Table 7
Performance of the multinomial classifier varying the number of features through PCA and autoencoders, respectively.
| Features | PCA | AE 3 layers | AE 9 layers | Accuracy | F1 | Accuracy | F1 | Accuracy | F1 |
| 70 | 0.960 | 0.961 | 0.960 | 0.961 | 0.960 | 0.961 | 60 | 0.978 | 0.978 | 0.917 | 0.919 | 0.989 | 0.989 | 50 | 0.983 | 0.983 | 0.901 | 0.900 | 0.979 | 0.980 | 35 | 0.980 | 0.980 | 0.912 | 0.912 | 0.976 | 0.977 | 25 | 0.930 | 0.931 | 0.903 | 0.904 | 0.928 | 0.928 |
|
|