CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 (test val)
6.64FID
6
CIFAR-10
91.6Accuracy
6
CIFAR10 1000 images (val)
88.67Clean Accuracy
6
CIFAR10 (test)
1.03Median ||δ||p
6
CIFAR-10
81.17Anat
6
CIFAR-10 (test)
0.94Accuracy
6
CIFAR-100 OOD (test)
77.01Test Accuracy
6
CIFAR-10 (val)
92.19Accuracy
6
CIFAR-10 clean data SIG attack 10% (train)
93.65Accuracy
6
CIFAR-10 10% clean data, CLB attack (train)
93.78Accuracy
6
CIFAR-10 IAB-one attack 10% (train)
93.89Accuracy
6
CIFAR-10 10% clean data Blend attack (train)
94.82Accuracy
6
CIFAR-10 10% clean data, Badnets attack (train)
93.73Accuracy
6
CIFAR-10 (1% clean data (500 images))
93.73ACC (Badnets)
6
CIFAR-10 1% (500 images) (train)
93.73Badnets Defense ACC
6
CIFAR-100-LT STEP profile (test)
55.19Balanced Error Rate
6
CIFAR-10-LT STEP-100 profile (test)
27.57Balanced Error
6
CIFAR-10
30Top-1 Error Rate
6
Imbalanced CIFAR-100 step, ratio 100 (val)
51.73Top-1 Error
6
CIFAR-100 32x32 (train test)
19.72FID
6
CIFAR-10 vs CIFAR-10.1 1.0 (test)
0.744Mean Rejection Rate
6
CIFAR-10 Symmetry-80% (test)
27.78Accuracy
6
CIFAR to STL
83.1Accuracy
6
CIFAR-10 (val)
95.01Accuracy
6
CIFAR-10 Symmetric noise 0.8 manually corrupted (test)
52.27Accuracy
6