Research Article
[Retracted] Unsupervised Anomaly Detection Based on Deep Autoencoding and Clustering
Table 3
Experimental results on Arrhythmia dataset.
| | Arrhythmia | Precision | F1 | Time (s) |
| K-Means | 0.1899 | 0.2776 | 0.7734 | PCA + K-Means | 0.1538 | 0.2424 | 0.6006 | ICA + K-Means | 0.1132 | 0.1846 | 0.1842 | MDS + K-Means | 0.1333 | 0.2143 | 0.3223 | DAE + K-Means | 0.3545 | 0.4432 | 0.7923 |
| DBSCAN | 0.2457 | 0.3772 | 0.0537 | PCA + DBSCAN | 0.2000 | 0.3056 | 1.0822 | ICA + DBSCAN | 0.1400 | 0.2295 | 0.6780 | MDS + DBSCAN | 0.1149 | 0.2000 | 0.8024 | DAE + DBSCAN | 0.5000 | 0.5217 | 0.5873 |
| Mean-Shift | 0.5000 | 0.4375 | 0.2449 | PCA + Mean-Shift | 0.5000 | 0.4615 | 0.8521 | ICA + Mean-Shift | 0.6667 | 0.4000 | 0.3851 | MDS + Mean-Shift | 0.5833 | 0.5000 | 0.5987 | DAE + Mean-Shift | 0.8158 | 0.5962 | 0.2980 |
|
|