Image Classification on CIFAR-100 (test) (ε and CI Metrics)
78.93AccuracyLeRaC
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LeRaCBackbone=CvT-13 (pre-trained)2022.05 | 78.93 | — | — | |
| LSCLBackbone=CvT-13 (pre-trained)2022.05 | 78.63 | — | — | |
| LCDnet-CLBackbone=CvT-13 (pre-trained)2022.05 | 78.57 | — | — | |
| EfficientTrainBackbone=CvT-13 (pre-trained)2022.05 | 78.2 | — | — | |
| Self-taughtBackbone=CvT-13 (pre-trained)2022.05 | 77.95 | — | — | |
| conventionalBackbone=CvT-13 (pre-trained)2022.05 | 77.8 | — | — | |
| CLIPBackbone=CvT-13 (pre-trained)2022.05 | 76.18 | — | — | |
| ALIBIAccuracy level=Very High, Backbone=Wide-ResNet182021.06 | 75.3 | 8.1 | — | |
| EfficientTrainBackbone=ResNet-182022.05 | 72.83 | — | — | |
| CBSBackbone=ResNet-182022.05 | 72.8 | — | — | |
| LeRaCBackbone=ResNet-182022.05 | 72.72 | — | — | |
| LSCLBackbone=Wide-ResNet-502022.05 | 72.59 | — | — | |
| Self-taughtBackbone=ResNet-182022.05 | 72.1 | — | — | |
| conventionalBackbone=ResNet-182022.05 | 71.7 | — | — | |
| ALIBIAccuracy level=High, Backbone=Wide-ResNet182021.06 | 71.4 | 6.3 | — | |
| LCDnet-CLBackbone=ResNet-182022.05 | 71.06 | — | — | |
| DSTBackbone=ResNet50, Pre-training=supervised2022.02 | 70.4 | — | — | |
| CLIPBackbone=ResNet-182022.05 | 70.03 | — | — | |
| PATE-FMAccuracy level=High, Backbone=Wide-ResNet18, δ=10^-52021.06 | 69.9 | 715 | — | |
| LeRaCBackbone=Wide-ResNet-502022.05 | 69.38 | — | — | |
| EfficientTrainBackbone=Wide-ResNet-502022.05 | 69.14 | — | — | |
| LCDnet-CLBackbone=Wide-ResNet-502022.05 | 68.85 | — | — | |
| Self-taughtBackbone=Wide-ResNet-502022.05 | 68.48 | — | — | |
| LSCLBackbone=ResNet-182022.05 | 68.42 | — | — | |
| conventionalBackbone=Wide-ResNet-502022.05 | 68.14 | — | — | |
| CLIPBackbone=Wide-ResNet-502022.05 | 68.13 | — | — | |
| CBSBackbone=Wide-ResNet-502022.05 | 65.73 | — | — | |
| ResNet18-Entropy#Samples=14000, Backbone=ResNet-182022.10 | 63.78 | — | — | |
| ResNet18-core-set#Samples=14000, Backbone=ResNet-182022.10 | 63.61 | — | — | |
| FlexMatchBackbone=ResNet50, Pre-training=supervised, Method Category=Dynamic Thresholding2022.02 | 63.4 | — | — | |
| FR-ResNet18-Entropy#Samples=14000, Backbone=ResNet-182022.10 | 63.27 | — | — | |
| CBSBackbone=CvT-13 (pre-trained)2022.05 | 62.35 | — | — | |
| FR-ResNet18-Entropy#Samples=13000, Backbone=ResNet-182022.10 | 62.01 | — | — | |
| LLAL#Samples=14000, Backbone=ResNet-182022.10 | 61.93 | — | — | |
| ResNet18-core-set#Samples=13000, Backbone=ResNet-182022.10 | 61.67 | — | — | |
| ResNet18-Entropy#Samples=13000, Backbone=ResNet-182022.10 | 61.39 | — | — | |
| LLAL#Samples=13000, Backbone=ResNet-182022.10 | 60.78 | — | — | |
| FR-ResNet18-Entropy#Samples=12000, Backbone=ResNet-182022.10 | 60.64 | — | — | |
| ResNet18-core-set#Samples=12000, Backbone=ResNet-182022.10 | 60.43 | — | — | |
| ResNet18-Entropy#Samples=12000, Backbone=ResNet-182022.10 | 59.9 | — | — | |
| FR-ResNet18-Entropy#Samples=11000, Backbone=ResNet-182022.10 | 59.73 | — | — | |
| LLAL#Samples=12000, Backbone=ResNet-182022.10 | 59.34 | — | — | |
| ResNet18-core-set#Samples=11000, Backbone=ResNet-182022.10 | 58.8 | — | — | |
| ResNet18-Entropy#Samples=11000, Backbone=ResNet-182022.10 | 58 | — | — | |
| LLAL#Samples=11000, Backbone=ResNet-182022.10 | 57.91 | — | — | |
| FR-ResNet18-Entropy#Samples=10000, Backbone=ResNet-182022.10 | 57.61 | — | — | |
| ResNet18-core-set#Samples=10000, Backbone=ResNet-182022.10 | 56.35 | — | — | |
| LLAL#Samples=10000, Backbone=ResNet-182022.10 | 56.18 | — | — | |
| FR-ResNet18-Entropy#Samples=9000, Backbone=ResNet-182022.10 | 56.02 | — | — | |
| ResNet18-Entropy#Samples=10000, Backbone=ResNet-182022.10 | 55.96 | — | — | |
| DashBackbone=ResNet50, Pre-training=supervised, Method Category=Dynamic Thresholding2022.02 | 55.4 | — | — | |
| ResNet18-core-set#Samples=9000, Backbone=ResNet-182022.10 | 54.57 | — | — | |
| Co-trainingBackbone=ResNet50, Pre-training=supervised, Method Category=Multi-head Training2022.02 | 54.4 | — | — | |
| MT Tri-trainingBackbone=ResNet50, Pre-training=supervised, Method Category=Multi-head Training2022.02 | 54.4 | — | — | |
| LLAL#Samples=9000, Backbone=ResNet-182022.10 | 54.2 | — | — | |
| FR-ResNet18-Entropy#Samples=8000, Backbone=ResNet-182022.10 | 53.8 | — | — | |
| ResNet18-Entropy#Samples=9000, Backbone=ResNet-182022.10 | 53.7 | — | — | |
| FixMatchBackbone=ResNet50, Pre-training=supervised2022.02 | 53.1 | — | — | |
| LLAL#Samples=8000, Backbone=ResNet-182022.10 | 51.62 | — | — | |
| ALIBIAccuracy level=Medium, Backbone=Wide-ResNet182021.06 | 51.6 | 3 | — | |
| FR-ResNet18-Entropy#Samples=7000, Backbone=ResNet-182022.10 | 51.21 | — | — | |
| ResNet18-core-set#Samples=8000, Backbone=ResNet-182022.10 | 50.26 | — | — | |
| ResNet18-Entropy#Samples=8000, Backbone=ResNet-182022.10 | 50.05 | — | — | |
| PATE-FMAccuracy level=Medium, Backbone=Wide-ResNet18, δ=10^-52021.06 | 50 | 16 | — | |
| FR-ResNet18-Entropy#Samples=6000, Backbone=ResNet-182022.10 | 48.91 | — | — | |
| LLAL#Samples=7000, Backbone=ResNet-182022.10 | 48.26 | — | — | |
| BaselineBackbone=ResNet50, Pre-training=supervised2022.02 | 48.2 | — | — | |
| ResNet18-core-set#Samples=7000, Backbone=ResNet-182022.10 | 48.1 | — | — | |
| ResNet18-Entropy#Samples=7000, Backbone=ResNet-182022.10 | 45.99 | — | — | |
| FR-ResNet18-Entropy#Samples=5000, Backbone=ResNet-18, Acquisition Strategy=Maximum Entropy2022.10 | 44.22 | — | — | |
| LLAL#Samples=6000, Backbone=ResNet-182022.10 | 43.73 | — | — | |
| ResNet18-core-set#Samples=6000, Backbone=ResNet-182022.10 | 43.62 | — | — | |
| ResNet18-Entropy#Samples=6000, Backbone=ResNet-182022.10 | 41.95 | — | — | |
| ResNet18-Entropy#Samples=5000, Backbone=ResNet-18, Acquisition Strategy=Maximum Entropy2022.10 | 38.73 | — | — | |
| ResNet18-core-set#Samples=5000, Backbone=ResNet-18, Acquisition Strategy=Core-set2022.10 | 38.6 | — | — | |
| LLAL#Samples=5000, Backbone=ResNet-182022.10 | 37.36 | — | — | |
| ALIBIAccuracy level=Low, Backbone=Wide-ResNet182021.06 | 31.4 | 2 | — | |
| PATE-FMAccuracy level=Low, Backbone=Wide-ResNet18, δ=10^-52021.06 | 30.5 | 7.9 | — |