CIFAR-10
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 Superpixels
80.19Accuracy
13
CIFAR-10 5%-200 devices
83.14Accuracy (Dirichlet α=0.5)
13
CIFAR-10 (test)
94.21Accuracy
13
CIFAR-10 C
47.8Accuracy
13
CIFAR-10 (Dirichlet (0.6), 500 clients, 2% participation)
78.49Accuracy (500 Rounds)
13
CIFAR-10 500 clients, 1% participation Dirichlet 0.3 (train test)
64.55Accuracy (500 Rounds)
13
CIFAR-10 LT (ρ = 0.01)
64.8AUPR (SVHN)
13
CIFAR-10-LT ρ = 0.01
68.81Accuracy
13
CIFAR-10 Distribution II: Quantity Skew (test)
45.16Accuracy
13
CIFAR-10 Distribution I: Label Scarcity (test)
0.613Accuracy
13
CIFAR-10 imbalanced µ=0.9, ρ=2 (test)
96.7Badnets TPR
13
CIFAR-10 imbalanced mu=0.9, rho=200 (train test)
83.3Badnets TPR
13
CIFAR-10 FedAvg with 50 clients, beta=5 non-IID (test)
82.42Accuracy
13
CIFAR-10 FedAvg with 50 clients, beta=0.1 non-IID (test)
68.44Accuracy
13
CIFAR-10 (test)
79.7Accuracy (beta=0.1)
13
CIFAR-10 non-IID (test)
85.97Clean MA
13
CIFAR-10 2009 (test)
1.79FID
13
CIFAR-10 vs SVHN
99.9ROCAUC
13
CIFAR-10.1 (test)
10.4Test Error
13
CIFAR-10 (test)
92.03Accuracy (10-class)
12
CIFAR-10 (test)
92.22Top-1 Accuracy (Epoch 50)
12
CIFAR-10 (test)
71.31Deferral Ratio (Expert 1)
12
CIFAR-10
0.106Deferral Loss
12
CIFAR-10-C mixture of 15 corruptions (val)
61Gaussian Noise Accuracy
12
CIFAR-10 10 samples per class
97.1Top-1 Accuracy
12