Research Article
Correlation-Based Anomaly Detection Method for Multi-sensor System
Table 7
Classification ability of anomaly detection in terms of AUC score, on three datasets.
| Method | Dataset 1 | Dataset 2 | Dataset 3 | Mean | Standard deviation | Mean | Standard deviation | Mean | Standard deviation |
| Logistic regression | 0.90 | 0.006 | 0.98 | 0.008 | 0.95 | 0.008 | Decision tree | 0.90 | 0.006 | 0.89 | 0.051 | 0.96 | 0.012 | Naïve Bayes | 0.93 | 0.007 | 0.96 | 0.010 | 0.84 | 0.025 | QDA | 0.92 | 0.009 | 0.91 | 0.030 | 0.92 | 0.020 | SVM | 0.75 | 0.012 | 0.81 | 0.022 | 0.68 | 0.042 | CNN | 0.96 | 0.004 | 0.96 | 0.007 | 0.93 | 0.006 | Proposed method | 0.96 | 0.005 | 0.98 | 0.009 | 0.95 | 0.005 |
|
|
Bold font is to highlight the results of the proposed method.
|