Image Classification on COVIDx CRX-2 (500 samples)
91.56Mean AccuracyLS+WRN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LS+WRNInit.=TF, Parameterization=Equivariant2021.07 | 91.56 | 0.52 | |
| LS+WRNInit.=TF, Parameterization=Canonical2021.07 | 90.5 | 0.7 | |
| S+WRNInit.=TF, Parameterization=Fixed2021.07 | 89.05 | 0.59 | |
| LS+LLInit.=TF, Parameterization=Canonical2021.07 | 88.5 | 0.71 | |
| LS+WRNInit.=Rand, Parameterization=Canonical2021.07 | 88.08 | 0.7 | |
| LS+LLInit.=TF, Parameterization=Equivariant2021.07 | 87.56 | 0.94 | |
| LS+WRNInit.=Rand, Parameterization=Equivariant2021.07 | 87.44 | 1.17 | |
| S+LLInit.=TF, Parameterization=Fixed2021.07 | 87.2 | 0.77 | |
| S+WRNInit.=Rand, Parameterization=Fixed2021.07 | 86.68 | 0.59 | |
| LS+LLInit.=Rand, Parameterization=Canonical2021.07 | 86.45 | 0.95 | |
| ResNet-502021.07 | 86.45 | 0.51 | |
| WRN-162021.07 | 85.95 | 2.04 | |
| LS+LLInit.=Rand, Parameterization=Equivariant2021.07 | 85.64 | 1.38 | |
| S+LLInit.=Rand, Parameterization=Fixed2021.07 | 84.13 | 0.91 |