Out-of-distribution Detection on CELEBA vs FOOD-101 (test)
0.96AUROCDE
Evaluation Results
| Method | Links | |
|---|---|---|
| DEMemory budget=10p2024.09 | 0.96 | |
| Sketched Lanczos UncertaintyMemory budget=10p2024.09 | 0.95 | |
| Sketched Lanczos UncertaintyModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.95 | |
| LLAMemory budget=10p2024.09 | 0.94 | |
| LEMemory budget=10p2024.09 | 0.94 | |
| SCODMemory budget=10p2024.09 | 0.94 | |
| SCODModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.94 | |
| LE-HMemory budget=10p2024.09 | 0.93 | |
| LLAModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.93 | |
| LEModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.93 | |
| LE-HModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.91 | |
| Deep EnsembleModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.88 | |
| LLA-DMemory budget=10p2024.09 | 0.8 | |
| LLA-DModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.8 | |
| SWAGMemory budget=10p2024.09 | 0.78 | |
| SWAGModel Architecture=VisualAttentionNet, Parameters (p)=4M, Memory Budget=3p2024.09 | 0.69 |