Fine-grained Classification on Stanford Dogs 40% symmetric noise (test)
79.12Best AccuracyDivideMix + SNSCL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DivideMix + SNSCLSNSCL=true2023.03 | 79.12 | 78.91 | |
| DivideMixSNSCL=false2023.03 | 77.93 | 76.28 | |
| MLC + SNSCLSNSCL=true2023.03 | 76.49 | 75.96 | |
| SYM + SNSCLSNSCL=true2023.03 | 76.28 | 76.25 | |
| Cross-Entropy + SNSCLSNSCL=true2023.03 | 75.27 | 75 | |
| MW-Net + SNSCLSNSCL=true2023.03 | 74.92 | 74.38 | |
| Conf. Penalty + SNSCLSNSCL=true2023.03 | 74.72 | 74.49 | |
| Label Smooth + SNSCLSNSCL=true2023.03 | 74.64 | 74.6 | |
| GCE + SNSCLSNSCL=true2023.03 | 71.68 | 70.62 | |
| Label SmoothSNSCL=false2023.03 | 70.22 | 50.97 | |
| MLCSNSCL=false2023.03 | 69.44 | 66.28 | |
| Cross-EntropySNSCL=false2023.03 | 69.2 | 50.45 | |
| Conf. PenaltySNSCL=false2023.03 | 68.69 | 52.98 | |
| MW-NetSNSCL=false2023.03 | 68.14 | 65.17 | |
| SYMSNSCL=false2023.03 | 65.76 | 46.99 | |
| JoCoR + SNSCLSNSCL=true2023.03 | 63.42 | 62.84 | |
| GCESNSCL=false2023.03 | 61.47 | 60.32 | |
| Co-teaching + SNSCLSNSCL=true2023.03 | 60.71 | 58.84 | |
| JoCoRSNSCL=false2023.03 | 49.62 | 48.62 | |
| Co-teachingSNSCL=false2023.03 | 49.15 | 48.92 |