Table 1.
In the ablation investigation, the performance of a basic Transformer Encoder and the suggested Multi-head Network were assessed using the given datasets.
| Datasets | Model | Accuracy (%) | F1 Score | Training Time/Hour | Testing Time/sec |
|---|---|---|---|---|---|
| Indian Pines 5% | Simple Encoder Transformer | 86.9 | 84.1 | 4.6 | 54 |
| A proposed multi-head network | 89.2 | 87.3 | 4.3 | 44 | |
| Pavia University 1% | Simple Encoder Transformer | 88.2 | 85.3 | 6.2 | 69 |
| A proposed multi-head network | 92.3 | 88.4 | 4.9 | 64 | |
| Salinas 1% | Simple Encoder Transformer | 87.5 | 84.5 | 6.6 | 64 |
| A proposed multi-head network | 90.7 | 87.1 | 5.9 | 54 | |
| KSA 10% | Simple Encoder Transformer | 84.3 | 81.2 | 5.7 | 59 |
| A proposed multi-head network | 93.2 | 90.1 | 6.4 | 59 |