CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 class-wise
99.71Accuracy Difference (Df)
8
CIFAR-100 (ID)
61Near-domain AUROC
8
CIFAR-10 (ID)
86Near-AUROC
8
CIFAR-100 10% Random Forgetting VGG16-BN
99.51Retention Accuracy (RA)
8
CIFAR-10 Mix noise 10% clean 90% noisy (train)
16.2FID
8
CIFAR-10 Poisson noise 10% clean 90% noisy (train)
14.7FID
8
CIFAR-10 Gaussian noise 10% clean, 90% noisy (train)
9.8FID
8
CIFAR-10 50% Random Forgetting Retain set D_r (train)
99.3Accuracy
8
CIFAR-10 30% Random Forgetting Retain D_r^train (train)
99.2Accuracy
8
CIFAR-10 30% class-wise forgetting (train test)
99.3Utility Retention D^train_r
8
CIFAR-100
86.67MP
8
CIFAR-10
90.7Accuracy (ACC)
8
CIFAR-100
82.09AUC
8
CIFAR-FS (meta-test)
41.91Best Accuracy (5-way 1-shot)
8
CIFAR 16x16
40.279FID (5k)
8
CIFAR16x16
0.019MSE
8
CIFAR-10 standard (test)
51.12Top-1 Accuracy
8
CIFAR-10 poisoned (train)
100Recall@50
8
CIFAR-100 ||δ||∞ ≤ 4.0/255 (test)
0.829AUC
8
CIFAR-10 ||δ||∞ ≤ 4.0/255 (test)
0.807AUC
8
CIFAR-100
0.788AUC
8
CIFAR-10
0.787AUC
8
Cifar10-LT exponential
83.66Accuracy
8
CIFAR-10
78.8Acc (Dirichlet, alpha=0.1)
8
CIFAR-10
93.69Accuracy
8