Robust Image Classification on CIFAR-100 RobustBench standard (val)
73.85Clean AccuracyIKL-KD
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| IKL-KDGenerated Data=w/, augmentation strategy=Generated data2023.05 | 73.85 | 39.18 | 0 | |
| Previous best resultsGenerated Data=w/, augmentation strategy=Generated data2023.05 | 72.58 | 38.83 | — | |
| Previous best resultsGenerated Data=w/o, augmentation strategy=Autoaug2023.05 | 68.75 | 31.85 | — | |
| IKL-KDGenerated Data=w/o, augmentation strategy=Autoaug2023.05 | 66.08 | 32.53 | 33.3 | |
| IKL-KDGenerated Data=w/o, augmentation strategy=Basic2023.05 | 65.76 | 31.91 | 33.3 | |
| Previous best resultsGenerated Data=w/o, augmentation strategy=Basic2023.05 | 62.99 | 31.2 | — |