Image Classification on CIFAR-100 (test) (Instance Accuracy Metrics)
70.03Accuracy (ins-20%)Co-learning+HCR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Co-learning+HCRBackbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 70.03 | 66.89 | 62.91 | |
| Co-learningBackbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 69.42 | 65.45 | 60.4 | |
| Standard CEBackbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 55.45 | 48.77 | 41.3 | |
| Co-teachingBackbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 55.16 | 45.24 | 34.64 | |
| JoCoRBackbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 54.21 | 45.03 | 34.08 | |
| DecouplingBackbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 52.2 | 45.32 | 36.33 | |
| Co-teaching+Backbone=ResNet-18, Number of trials=5, Evaluation Protocol=Average test accuracy over last 10 epochs2022.06 | 50.37 | 40.73 | 32.15 |