CIFAR
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CIFAR-10 Deep
5.69Verification Time (s)
12
CIFAR-10 Wide
78.32Verification Time (s)
12
CIFAR-10 Base
118.23Verification Time (s)
12
CIFAR-100 5-way 5-shot
74.45Accuracy (Session 1)
12
CIFAR-10
7.76FID
12
CIFAR-10 0.1% clean data (50 images) (train)
93.73ACC (Badnets)
12
CIFAR-10 Step (rho=50) (test)
17.9Top-1 Error Rate
12
CIFAR-100 (test)
71.13Recall@1
12
CIFAR-100 Imbalanced (long-tailed, ratio 10) (val)
37.54Top-1 Error
12
CIFAR-10 (test)
99.6Sparsity
12
CIFAR-100 (test)
77.7Top-1 Accuracy
12
CIFAR-100 super-class
89.6AUROC
12
CIFAR-10 Split
98.3Average Accuracy
12
CIFAR-10 known attack 1.0 (test)
99.94AUROC (FGSM)
12
CIFAR 100 10 different calib. (test)
42.75Mean Prediction Set Size
12
CIFAR 10 different calib (test)
2.22Mean Prediction Set Size
12
CIFAR 100 (test)
41.18Mean Prediction Set Size
12
CIFAR 10 (test)
2.04Mean Prediction Set Size
12
CIFAR-10 (test)
100Solution Count
12
CIFAR-100 (full)
83.05Accuracy
12
CIFAR-10
48.39System Accuracy
11
CIFAR10 unlearning (Du)
99.93Accuracy
11
CIFAR-100 Heterogeneous (test)
1Communication Overhead Factor
11
CIFAR-10
0.66Aggregated Local-Global Score
11
CIFAR-10 10% random data forgetting
56.12Utility Accuracy (UA)
11