Image Classification on CIFAR100 (test) (Acc, CI, DBI)
86.65AccuracyFoCA ViT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FoCA ViTBackbone=ViT2026.06 | 86.65 | 75.5 | 1.983 | |
| ViT CBMBackbone=ViT2026.06 | 84.49 | 77.4 | 2.237 | |
| FoCA CBM-NBackbone=ResNet2026.06 | 82.41 | 68.8 | 2.177 | |
| FoCA CBMBackbone=ResNet2026.06 | 79.47 | 62.2 | 1.855 | |
| SCBMBackbone=ResNet2026.06 | 79.13 | 70.5 | 2.114 | |
| ProbCBMBackbone=ResNet2026.06 | 78.86 | 73.2 | 2.401 | |
| CEMBackbone=ResNet2026.06 | 77.32 | 81.1 | 2.497 | |
| Vanilla CBMBackbone=ResNet2026.06 | 76.63 | 71.2 | 2.238 | |
| MLPCBMBackbone=ResNet2026.06 | 76.17 | 72.2 | 2.276 | |
| LaBoBackbone=ResNet2026.06 | 65.23 | 90.7 | 2.519 | |
| LFCBMBackbone=ResNet2026.06 | 65.13 | 90.7 | 2.519 | |
| CF-CBMBackbone=ResNet2026.06 | 60.02 | 90.7 | 2.519 | |
| HybridCBMBackbone=ResNet2026.06 | 59.52 | 90.7 | 2.519 | |
| Posthoc CBMBackbone=ResNet2026.06 | 52 | 85.1 | 2.779 |