CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 10 images/class (test)
67.5Accuracy
15
CIFAR-10 one vs. rest 80MTI as OE
99.6Mean AUC
15
CIFAR10 beta=100 LT (test)
82.6Top-1 Acc
15
CIFAR-10
2.65NLL
15
CIFAR-10 transferred from STL-10 (test)
49.2Accuracy
15
CIFAR-10 (test)
96.3Accuracy (Symmetric 20% noise)
15
CIFAR-10 PGD-100, l_inf, 16/255 (test)
48.02TPR-95
15
CIFAR-10 PGD-100, l_inf, 8/255 (test)
58.52TPR-95
15
CIFAR-10 (test)
93.2Accuracy (Sym. Noise Rate 0.2)
15
CIFAR-10
94.56Accuracy
15
CIFAR-LT-100 Imbalance Factor 20 (test)
42.15Top-1 Error
15
CIFAR-LT-100 Imbalance Factor 200 (test)
56.91Top-1 Error
15
CIFAR10 4,000 labels (train)
12.41Error Rate
15
CIFAR-100 (test)
10.7Test Error Rate
15
CIFAR-100 (test)
68.75Symmetric Noise (η=0.2) Accuracy
15
CIFAR-10 (test)
100AUPR Succ
15
CIFAR-10
94.84InD ACC
14
CIFAR-10 AA
72.13Accuracy
14
CIFAR-100
0.91R^2
14
CIFAR-10H hard subset
64.8Accuracy
14
CIFAR-10
88.3Accuracy
14
CIFAR10
88.72Accuracy
14
CIFAR100
59.51Accuracy
14
CIFAR100 (test)
86.65Accuracy
14
CIFAR-10 (6-way sharded)
41.3Accuracy
14