Image Classification on CIFAR-10-N-LT Imbalance Ratio 100
79.36Accuracy (Noise 0.1)DivideMix + Balanced-Softmax
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DivideMix + Balanced-SoftmaxCheckpoint=Best2022.11 | 79.36 | 76.55 | 48.6 | 54.99 | 55.84 | 72.75 | |
| HARCheckpoint=Best2022.11 | 79.02 | 76.14 | 72.74 | 67.22 | 65 | 77.17 | |
| Dynamic LossCheckpoint=Best2022.11 | 78.96 | 76.64 | 76.17 | 70.37 | 70.26 | 80.55 | |
| Balanced-SoftmaxCheckpoint=Best2022.11 | 78.63 | 73.29 | 70.89 | 67.57 | 63.4 | 76.37 | |
| HARCheckpoint=Last2022.11 | 78.6 | 75.05 | 72.08 | 65.48 | 63.9 | 76.52 | |
| Dynamic LossCheckpoint=Last2022.11 | 77.8 | 76.31 | 74.1 | 69.64 | 67.45 | 79.73 | |
| DivideMix + Balanced-SoftmaxCheckpoint=Last2022.11 | 76.63 | 75.72 | 46.82 | 53.25 | 54.33 | 71.6 | |
| Balanced-SoftmaxCheckpoint=Last2022.11 | 76.14 | 69.37 | 62.03 | 52.79 | 46.44 | 69.66 | |
| CurveNetCheckpoint=Best2022.11 | 67.55 | 68.72 | 63.71 | 51.63 | 52.84 | 71.36 | |
| MOIT+Checkpoint=Best2022.11 | 66.92 | 62.16 | 60.49 | 37.85 | 39.19 | 68.48 | |
| MOIT+Checkpoint=Last2022.11 | 66.21 | 61.63 | 60.01 | 37.71 | 39.01 | 68.11 | |
| CurveNetCheckpoint=Last2022.11 | 65.77 | 66.21 | 62.37 | 48.71 | 51.85 | 69.36 | |
| ELR+Checkpoint=Best2022.11 | 65.73 | 62.12 | 56.42 | 52.26 | 45.73 | 70.34 | |
| ELR+Checkpoint=Last2022.11 | 64.76 | 61.41 | 56.42 | 51.31 | 44.05 | 69.31 | |
| Cross EntropyCheckpoint=Best2022.11 | 63.45 | 53.45 | 49.49 | 45.53 | 40.76 | 60.43 | |
| DivideMixCheckpoint=Last2022.11 | 63.08 | 61.14 | 24.31 | 24.13 | 20.09 | 58.19 | |
| DivideMixCheckpoint=Best2022.11 | 63.08 | 61.61 | 33.79 | 31.06 | 26.1 | 60.74 | |
| Cross EntropyCheckpoint=Last2022.11 | 59.99 | 50.56 | 44.81 | 37.13 | 30.02 | 52.73 | |
| FaMUSCheckpoint=Best2022.11 | 57.5 | 54.61 | 54.3 | 45.35 | 32.93 | 66.66 | |
| FaMUSCheckpoint=Last2022.11 | 55.46 | 51.67 | 49.4 | 38.21 | 26.69 | 61.53 |