| 模型 | Cifar-100 | Mini-ImageNet | ||||||
|
|
| By DLC-UG |
| By DLC-UG | ||||
| 分布 | Ran(0, 1) | Ran(−1, 1) | Ran(0, 1) | Ran(−1, 1) | Ran(0, 1) | Ran(−1, 1) | Ran(0, 1) | Ran(−1, 1) |
| VoVNet-57 [21] | 64.32% | 64.05% | 64.85% | 72.68% | 72.61% | 72.56% | 73.03% | 79.38% |
| VGG16 [22] | 45.70% | 46.12% | 46.45% | 56.59% | 43.67% | 43.52% | 44.39% | 53.32% |
| ResNeSt50 [23] | 66.35% | 66.73% | 66.91% | 74.69% | 69.82% | 69.87% | 70.21% | 77.67% |
| RepVGG [24] | 63.38% | 63.94% | 63.87% | 72.66% | 70.76% | 70.80% | 71.17% | 77.54% |
| ResNet [25] | 41.47% | 41.25% | 42.08% | 51.07% | 40.39% | 41.50% | 40.93% | 50.83% |
| DenseNet [26] | 54.89% | 54.79% | 55.71% | 65.22% | 50.80% | 51.04% | 51.45% | 61.60% |