Resolving Confusion on Lacuna-10 (test)
2.1Test ErrorSCRUB
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| SCRUBBackbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 2.1 | 0 | 97.5 | 4.17 | 0 | 0.33 | 0 | |
| retrainBackbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 2.13 | 0 | 99.83 | 2.5 | 0 | 0 | 0 | |
| EU-kBackbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 2.53 | 0.09 | 97.83 | 5.17 | 0.38 | 0.33 | 0.67 | |
| finetuneBackbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 3 | 0 | 74.5 | 6.5 | 0 | 9 | 0 | |
| NegGrad+Backbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 4.3 | 0.4 | 89.67 | 15.33 | 2.73 | 4.67 | 0 | |
| CF-kBackbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 5.27 | 0.11 | 33.33 | 18.5 | 0.76 | 31.33 | 3.33 | |
| originalBackbone=All-CNN, Confused classes=0,1, Number of samples=200-2002023.02 | 7.83 | 1.21 | 16 | 31.33 | 8.26 | 56.33 | 36.33 |