Image Classification on CIFAR100 (test) (Accuracy, F1, and Efficiency Metrics)
80.6AccuracyOrderDP
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| OrderDPBackbone=ResNet-50, Prune Ratio %=302026.06 | 80.6 | — | — | — | — | — | |
| Whole DatasetBackbone=ResNet-50, Prune Ratio %=02026.06 | 80.6 | — | — | — | — | — | |
| InfoBatchBackbone=ResNet-50, Prune Ratio %=302026.06 | 80.4 | — | — | — | — | — | |
| OrderDPBackbone=ResNet-50, Prune Ratio %=502026.06 | 79.8 | — | — | — | — | — | |
| InfoBatchBackbone=ResNet-50, Prune Ratio %=502026.06 | 78.6 | — | — | — | — | — | |
| OrderDPBackbone=ResNet-18, Prune Ratio %=302026.06 | 78.2 | — | — | — | — | — | |
| Whole DatasetBackbone=ResNet-18, Prune Ratio %=02026.06 | 78.2 | — | — | — | — | — | |
| InfoBatchBackbone=ResNet-18, Prune Ratio %=302026.06 | 78.1 | — | — | — | — | — | |
| UCBBackbone=ResNet-50, Prune Ratio %=302026.06 | 78 | — | — | — | — | — | |
| OrderDPBackbone=ResNet-18, Prune Ratio %=502026.06 | 77.9 | — | — | — | — | — | |
| Dynamic RandomBackbone=ResNet-50, Prune Ratio %=302026.06 | 77.9 | — | — | — | — | — | |
| OrderDPBackbone=ResNet-50, Prune Ratio %=702026.06 | 77.9 | — | — | — | — | — | |
| InfoBatchBackbone=ResNet-18, Prune Ratio %=502026.06 | 77.7 | — | — | — | — | — | |
| Dynamic RandomBackbone=ResNet-18, Prune Ratio %=302026.06 | 77.3 | — | — | — | — | — | |
| UCBBackbone=ResNet-18, Prune Ratio %=302026.06 | 77.3 | — | — | — | — | — | |
| ϵ-greedyBackbone=ResNet-50, Prune Ratio %=302026.06 | 77.2 | — | — | — | — | — | |
| OrderDPBackbone=ResNet-18, Prune Ratio %=702026.06 | 76.7 | — | — | — | — | — | |
| UCBBackbone=ResNet-50, Prune Ratio %=502026.06 | 76.5 | — | — | — | — | — | |
| ϵ-greedyBackbone=ResNet-18, Prune Ratio %=302026.06 | 76.4 | — | — | — | — | — | |
| InfoBatchBackbone=ResNet-50, Prune Ratio %=702026.06 | 76.4 | — | — | — | — | — | |
| ϵ-greedyBackbone=ResNet-50, Prune Ratio %=502026.06 | 76.3 | — | — | — | — | — | |
| Dynamic RandomBackbone=ResNet-50, Prune Ratio %=502026.06 | 76.1 | — | — | — | — | — | |
| InfoBatchBackbone=ResNet-18, Prune Ratio %=702026.06 | 75.9 | — | — | — | — | — | |
| Dynamic RandomBackbone=ResNet-18, Prune Ratio %=502026.06 | 75.3 | — | — | — | — | — | |
| UCBBackbone=ResNet-18, Prune Ratio %=502026.06 | 75.3 | — | — | — | — | — | |
| ϵ-greedyBackbone=ResNet-18, Prune Ratio %=502026.06 | 74.8 | — | — | — | — | — | |
| UCBBackbone=ResNet-50, Prune Ratio %=702026.06 | 74.3 | — | — | — | — | — | |
| ϵ-greedyBackbone=ResNet-50, Prune Ratio %=702026.06 | 74.1 | — | — | — | — | — | |
| Dynamic RandomBackbone=ResNet-50, Prune Ratio %=702026.06 | 73.9 | — | — | — | — | — | |
| UCBBackbone=ResNet-18, Prune Ratio %=702026.06 | 73.2 | — | — | — | — | — | |
| ϵ-greedyBackbone=ResNet-18, Prune Ratio %=702026.06 | 72.9 | — | — | — | — | — | |
| Dynamic RandomBackbone=ResNet-18, Prune Ratio %=702026.06 | 72.8 | — | — | — | — | — | |
| FocalBackbone=PVT, gamma=2, alpha=0.252026.03 | 40.17 | 39.96 | 4.9609 | 12.57 | — | 50 | |
| FocalBackbone=PVT, gamma=3, alpha=0.252026.03 | 40.15 | 39.99 | 4.9567 | 13.64 | — | 50 | |
| FocalBackbone=PVT, gamma=2, alpha=0.52026.03 | 40.1 | 39.97 | 3.9182 | 12.71 | — | 50 | |
| Conf. PenaltyBackbone=PVT, beta=0.12026.03 | 39.99 | 39.77 | 5.2574 | 9.63 | — | 50 | |
| BaselineBackbone=PVT2026.03 | 39.94 | 39.7 | 3.67 | 9.73 | — | 50 | |
| Label SmoothingBackbone=PVT, epsilon=0.12026.03 | 38.94 | 38.88 | 2.6626 | 8.18 | — | 50 | |
| Conf. PenaltyBackbone=PVT, beta=0.22026.03 | 38.59 | 38.34 | 2.1718 | 7.49 | — | 50 | |
| Label SmoothingBackbone=PVT, epsilon=0.22026.03 | 38.47 | 38.51 | 2.2097 | 7.42 | — | 50 | |
| BaselineBackbone=CNN2026.03 | 37.95 | 37.95 | 1.18 | 45.9295 | 49.3333 | — | |
| ArcFaceBackbone=PVT, m=0.5, s=302026.03 | 37.28 | 37.72 | 2.6171 | 12.59 | — | 50 | |
| Bray-Curtis (Norm.)Backbone=CNN2026.03 | 32.29 | 31.82 | 0.8132 | 90.94 | 2.6667 | — | |
| Mahalanobis (Chol.)Backbone=CNN2026.03 | 29.27 | 29.21 | 0.727 | 34.1 | 50 | — | |
| EuclideanBackbone=PVT2026.03 | 28.64 | 29.45 | 6.4329 | 84.14 | — | 4 | |
| Cosine (Unst.)Backbone=CNN2026.03 | 26.02 | 26.67 | 2.1156 | 53.06 | 45 | — | |
| Cosine (Stable)Backbone=CNN2026.03 | 25.01 | 25.16 | 1.4263 | 52.16 | 45 | — | |
| EuclideanBackbone=CNN2026.03 | 24.13 | 24.31 | 1.2866 | 43.62 | 50 | — |