Image Classification on CIFAR-100 80% symmetric label noise
58.88AccuracyFedSIR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FedSIRalpha=0.5, # Clean Clients=62026.04 | 58.88 | — | — | |
| FedSIRalpha=0.3, # Clean Clients=72026.04 | 58.78 | — | — | |
| FedSIRalpha=2, # Clean Clients=52026.04 | 58.27 | — | — | |
| FedELCalpha=0.3, # Clean Clients=72026.04 | 58.17 | — | — | |
| FedNoRoalpha=2, # Clean Clients=52026.04 | 57.58 | — | — | |
| FedELCalpha=0.5, # Clean Clients=62026.04 | 57.3 | — | — | |
| FedELCalpha=2, # Clean Clients=52026.04 | 57.21 | — | — | |
| FedNoRoalpha=0.5, # Clean Clients=62026.04 | 57.1 | — | — | |
| FedNoRoalpha=0.3, # Clean Clients=72026.04 | 56.51 | — | — | |
| FedNedalpha=2, # Clean Clients=52026.04 | 56.43 | — | — | |
| FedNedalpha=0.3, # Clean Clients=72026.04 | 56.21 | — | — | |
| FedNedalpha=0.5, # Clean Clients=62026.04 | 55.7 | — | — | |
| FedProxalpha=0.3, # Clean Clients=72026.04 | 54.88 | — | — | |
| FedAvgalpha=0.3, # Clean Clients=72026.04 | 54.07 | — | — | |
| FedProxalpha=0.5, # Clean Clients=62026.04 | 53.58 | — | — | |
| FedAvgalpha=2, # Clean Clients=52026.04 | 53.49 | — | — | |
| FedProxalpha=2, # Clean Clients=52026.04 | 53.29 | — | — | |
| FedAvgalpha=0.5, # Clean Clients=62026.04 | 52.62 | — | — | |
| FedCorralpha=2, # Clean Clients=52026.04 | 51.22 | — | — | |
| RoFLalpha=2, # Clean Clients=52026.04 | 50.61 | — | — | |
| FedCorralpha=0.5, # Clean Clients=62026.04 | 49.65 | — | — | |
| FedCorralpha=0.3, # Clean Clients=72026.04 | 47.11 | — | — | |
| RoFLalpha=0.5, # Clean Clients=62026.04 | 44.92 | — | — | |
| FedLSRalpha=0.3, # Clean Clients=72026.04 | 44.74 | — | — | |
| FedLSRalpha=2, # Clean Clients=52026.04 | 43.73 | — | — | |
| FedLSRalpha=0.5, # Clean Clients=62026.04 | 43.48 | — | — | |
| RoFLalpha=0.3, # Clean Clients=72026.04 | 42.22 | — | — | |
| RHFLalpha=2, # Clean Clients=52026.04 | 19.18 | — | — | |
| RHFLalpha=0.5, # Clean Clients=62026.04 | 16.32 | — | — | |
| RHFLalpha=0.3, # Clean Clients=72026.04 | 15.51 | — | — | |
| C2D (DivideMix with SimCLR)Architecture=PreAct ResNet-182021.03 | — | 67.78 | 67.43 | |
| C2D (DivideMix with SimCLR)Architecture=ResNet-502021.03 | — | 71.65 | 71.53 | |
| C2D (ELR+ with SimCLR)Architecture=ResNet-342021.03 | — | 64.72 | 64.18 | |
| CE+mixup with SimCLRArchitecture=ResNet-342021.03 | — | 61.39 | 60.81 | |
| DivideMixArchitecture=PreAct ResNet-182021.03 | — | 61.6 | 61.3 | |
| ELR+Architecture=ResNet-342021.03 | — | — | 60.8 | |
| Meta-learningArchitecture=PreAct ResNet-322021.03 | — | 42.4 | 40.1 |