MNIST
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MNIST Label Noise 0.4 (test)
93.74Mean Accuracy
12
MNIST Label Noise 0.2 (test)
94.6Mean Accuracy
12
MNIST case (iv) (test)
90.94Accuracy
12
MNIST case (iii) (test)
95.16Accuracy
12
MNIST full (test)
99.23Natural Accuracy
12
MNIST (test)
99.2Clean Accuracy
12
MNIST (test)
69Clean Error
12
MNIST (test)
0.9Clean Error Rate
12
MNIST (0-4) vs {MNIST (5-9), EMNIST-Letters, Omniglot, Fashion-MNIST, SVHN, CIFAR10} (test)
0.999AUROC
12
MNIST (0-4) (test)
99.91Accuracy
12
MNIST (test)
99.05Acc (3 Classes/Client)
12
Rotated MNIST (target)
93.3Accuracy
12
MNIST 2-Class
0.946AUC
12
MNIST
0.93bpd
12
MNIST
2.1Avg Set Size
12
MNIST
41Median Score
11
MNIST (test)
90Median Performance
11
MNIST (MNI)
0.64Aggregated Local-Global Score
11
MNIST
1.96MAE (%)
11
MNIST mixture 784 dim (test)
1MSE
11
MNIST 50k (train)
0.75FID
11
Split-MNIST
2.6Average Forgetting
11
MNIST
99.35Accuracy
11
MNIST adapted from SVHN (test)
0.866R^2
11
MNIST (ID) -> KMNIST (OOD) (test)
99.98ID Accuracy
11