Research Article
Anomaly Detection Collaborating Adaptive CEEMDAN Feature Exploitation with Intelligent Optimizing Classification for IIoT Sparse Data
Table 1
Training and test classification accuracies of CEEMDAN and EEMD decompositions under different attack powers.
| Attack power | CEEMDAN | EEMD | Training accuracy | Test accuracy | Training accuracy | Test accuracy |
| 1 | 96.0% | 86.5% | 92.5% | 81.5% | 2 | 93.5% | 88.0% | 91.0% | 87.5% | 3 | 91.0% | 90.0% | 92.0% | 88.5% | 4 | 96.5% | 87.0% | 94.5% | 87.0% | 5 | 98.5% | 93.5% | 91.5% | 88.0% | Average | 95.10% | 89.00% | 92.30% | 86.50% |
|
|