Table 1.
Performance of three methods on simulated datasets
| Simulated ±60° | Simulated ±40° | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Dataset | SNR 0.1 | 0.03 | 0.01 | 0.003 | 0.001 | SNR 0.1 | 0.03 | 0.01 | 0.003 | 0.001 |
| - | - | - | - | - | - | - | - | - | - | |
| Template matching | 0.7013 | 0.4709 | 0.1496 | 0.0136 | 0.0032 | 0.5543 | 0.3336 | 0.0655 | 0.0062 | 0.0012 |
| 83.95% | 69.75% | 45.35% | 25.25% | 20.95% | 76.25% | 61.15% | 36.60% | 23.80% | 21.20% | |
| K = 5 | K = 4 | K = 5 | K = 5 | K = 3 | K = 6 | K = 5 | K = 3 | K = 3 | K = 3 | |
| Autoencoder | 0.3843 | 0.4539 | 0.3613 | 0.4915 | 0.3881 | 0.5227 | 0.3470 | 0.3735 | 0.3878 | 0.3874 |
| 56.75% | - | 53.45% | 64.80% | - | - | 53.35% | - | - | - | |
| K = 5 | K = 5 | K = 5 | K = 5 | K = 5 | K = 5 | K = 5 | K = 5 | K = 6 | K = 6 | |
| DISCA | 0.9878 | 0.9373 | 0.8746 | 0.8712 | 0.8719 | 0.9568 | 0.8020 | 0.8344 | 0.8366 | 0.8323 |
| 99.70% | 97.80% | 94.80% | 94.25% | 94.50% | 98.70% | 90.35% | 91.80% | - | - | |
In each cell, the first row denotes the estimated K for unsupervised methods. The second row denotes the homogeneity score compared to the ground truth. The third row denotes prediction accuracy.