Knowledge Distillation on CIFAR-100 + Tiny-ImageNet
0.855AccuracyMDA
Evaluation results
| Method | Links | |
|---|---|---|
| MDABackbone=VGG-16, Category=0+02022.03 | 0.855 | |
| Base (Source models average)Backbone=VGG-16, Category=0+02022.03 | 0.83 | |
| Knowledge Amalgamating (KA)Backbone=VGG-16, Category=0+02022.03 | 0.8131 | |
| Base (Source models average)Backbone=VGG-16, Category=20-69 + 70-1792022.03 | 0.7148 | |
| Base (Source models average)Backbone=VGG-16, Category=0-19 + 0-692022.03 | 0.6986 | |
| Knowledge Amalgamating (KA)Backbone=VGG-16, Category=0-19 + 0-692022.03 | 0.6824 | |
| Knowledge Amalgamating (KA)Backbone=VGG-16, Category=20-69 + 70-1792022.03 | 0.6752 | |
| Base (Source models average)Backbone=VGG-16, Category=0-99 + 0-1992022.03 | 0.5597 | |
| Knowledge Amalgamating (KA)Backbone=VGG-16, Category=0-99 + 0-1992022.03 | 0.5274 | |
| MDABackbone=VGG-16, Category=0-19 + 0-692022.03 | 0.5066 | |
| MDABackbone=VGG-16, Category=20-69 + 70-1792022.03 | 0.5008 | |
| MDABackbone=VGG-16, Category=0-99 + 0-1992022.03 | 0.4306 |