Image Classification on Caltech-UCSD Birds (CUB-200-2011) (test)
93.29AccuracyMoCo V2 Anisotropic
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MoCo V2 Anisotropictraining_mode=SSL, texture_suppression=anisotropic2020.11 | 93.29 | — | — | |
| Neural NetworkModel Transparency=Black-box, Backbone=ResNet102021.06 | 93.1 | — | — | |
| Entropy netModel Transparency=White-box2021.06 | 92.95 | — | — | |
| MoCo V2training_mode=SSL2020.11 | 92.13 | — | — | |
| ψ netModel Transparency=White-box2021.06 | 91.92 | — | — | |
| Random ForestModel Transparency=Black-box2021.06 | 91.88 | — | — | |
| BRLModel Transparency=White-box2021.06 | 90.79 | — | — | |
| Supervised (Reproduced)training_mode=supervised2020.11 | 90.3 | — | — | |
| Coop-CBMLoss=Concept Orthogonal Loss (COL)2023.11 | 84.1 | — | — | |
| Coop-CBM2023.11 | 83.6 | — | — | |
| CEM2023.11 | 82.5 | — | — | |
| StandardModel type=No concepts2023.11 | 82.3 | — | — | |
| TreeModel Transparency=White-box, Constraint=Height limited2021.06 | 81.62 | — | — | |
| CBM-AR2023.11 | 81.6 | — | — | |
| Joint CBM2023.11 | 80.1 | — | — | |
| Sequential CBM2023.11 | 76.3 | — | — | |
| Independent CBM2023.11 | 76 | — | — | |
| PSCBMtraining_interventions=false2025.10 | 68.3983 | 96.8 | 740 | |
| PSCBMitraining_interventions=true2025.10 | 68.197 | 97.04 | 14,084 | |
| CBM2025.10 | 67.3973 | 95.51 | 7,204 | |
| SCBMcovariance_learning=global2025.10 | 65.5173 | 96.71 | 8,134 | |
| Finetuning (Cross-Entropy)Backbone=ResNet-18, Evaluation Protocol=Observed Accuracy2022.03 | 54.5 | — | — | |
| Finetuning (SupCon)Backbone=ResNet-18, Evaluation Protocol=Observed Accuracy2022.03 | 50.4 | — | — | |
| EWCBackbone=ResNet-18, Evaluation Protocol=Observed Accuracy, lambda=0.5k2022.03 | 47.8 | — | — | |
| LwFBackbone=ResNet-18, Evaluation Protocol=Observed Accuracy2022.03 | 43.1 | — | — | |
| EWCBackbone=ResNet-18, Evaluation Protocol=Observed Accuracy, lambda=8k2022.03 | 38.3 | — | — |