Robust Image Classification on CIFAR10 (test) (Flip Noise)
92.95Accuracy (0% Flip Noise)MWNet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MWNetNoise Type=Flip noise, Number of runs=52021.10 | 92.95 | 89.93 | 85.55 | |
| GDWNoise Type=Flip noise, Number of runs=52021.10 | 92.94 | 91.05 | 87.7 | |
| Fine-tuningNoise Type=Flip noise, Number of runs=52021.10 | 92.77 | 90.15 | 81.53 | |
| BaseModelNoise Type=Flip noise, Number of runs=52021.10 | 92.73 | 90.14 | 81.2 | |
| INSWNoise Type=Flip noise, Number of runs=52021.10 | 92.7 | 89.9 | 80.09 | |
| Soft-labelNoise Type=Flip noise, Number of runs=52021.10 | 92.63 | 90.17 | 85.52 | |
| Gen-labelNoise Type=Flip noise, Number of runs=52021.10 | 92.56 | 90.18 | 80.93 | |
| Co-teachingNoise Type=Flip noise, Number of runs=52021.10 | 91.54 | 89.27 | 69.77 | |
| GLCNoise Type=Flip noise, Number of runs=52021.10 | 90.85 | 90.22 | 89.74 | |
| L2RWNoise Type=Flip noise, Number of runs=52021.10 | 89.7 | 88.21 | 82.9 |