CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 BLEND attack (test)
88.83Accuracy
6
CIFAR-10 (test)
42.74Type-I 70% Score
6
CIFAR-100 Type-III 35% Polynomial Margin Diminishing Noise 1.0 (test)
70.13Accuracy
6
CIFAR-100 Type-II 35% Polynomial Margin Diminishing Noise 1.0 (test)
69.35Accuracy
6
CIFAR-10 Type-III 35% Polynomial Margin Diminishing Noise 1.0 (test)
94.23Accuracy
6
CIFAR-10 first 100 examples (test)
18,638.14Avg Local Lipschitz Constant
6
CIFAR-10
29Failure Rate
6
CIFAR SGD first 1000 images (val)
-1.5Relative Bound
6
CIFAR SGD
1.5Time (s)
6
CIFAR-100 SMI scenario (test)
62.38Mf Score
6
CIFAR10-C vs CIFAR100 OOD (test)
0.748AUROC
6
CIFAR10-C vs SVHN OOD (test)
80.3AUROC
6
CIFAR-100 (50/50 seen/novel split)
57.5Accuracy
6
CIFAR-100 B50 25 steps (test)
81.74Average Accuracy (All Sessions)
6
CIFAR-10 RGB Trigger
29.8Detected L1 Norm
6
Sequential CIFAR-100 20 Tasks (test)
31.93Accuracy
6
CIFAR-100 Sequential
19.29Forgetting
6
CIFAR-100 IID p=0.4, τ=0.5
140Targeted Communication Cost
6
CIFAR-100 IID p=0.0, τ=0.0
95Targeted Communication Cost
6
CIFAR-100 40% asymmetric label noise
77.78Final Classification Accuracy
6
CIFAR-100 10000-label (train test)
23.04Median Error Rate
6
CIFAR-100 400-label (train test)
45.91Median Error Rate
6
CIFAR-10 4000-label (train test)
4.46Median Error Rate
6
CIFAR-10 400-label (train test)
5.12Median Error Rate
6
CIFAR-10 40-label (train/test)
5.19Median Error Rate
6