CIFAR-100
Benchmarks
TaskDatasetSOTA resultTrendResults
CIFAR-100 SCOOD framework (test)
60.37FPR95
9
CIFAR-100 beta=0.6 (Dirichlet)
49.08Accuracy
9
CIFAR-100 (test)
9,247.08Upload Throughput (x10^3 Scalars)
9
CIFAR-100 sr=90% (test)
64.37Clean Accuracy
9
CIFAR-100 non-IID (test)
63.33Clean Accuracy
9
CIFAR-100-C Severity 1-5
71.22Accuracy
9
CIFAR-100 Long-Tailed (test)
67.4Accuracy (Old)
9
CIFAR-100 ImageNet-1K transfer (test)
0.613SA
9
CIFAR-100-C 1.0 (test)
84.05AUROC
9
CIFAR-100 (test)
71.1Acc (Sym 20%)
9
Imbalanced CIFAR-100 Step
53Top-1 Error (rho=100)
9
CIFAR-100 Imbalanced (Long-Tailed)
55.45Top-1 Error (rho=100)
9
CIFAR-100
87.4Accuracy
8
CIFAR-100
93.5Accuracy
8
CIFAR-100 LT100
7.25Accuracy
8
CIFAR-100 Split
3.33Recovery Steps
8
Split-CIFAR-100 5 tasks x 20 classes (train)
43.1Average Accuracy (AA)
8
CIFAR-100 Dirichlet partitioning (test)
41.8Accuracy
8
CIFAR-100 sigma=0.01 (test)
51.58Test Accuracy
8
CIFAR-100-LT
5.14FID
8
CIFAR-100-C Severity 4 1 (test)
7.2ECE
8
CIFAR-100-C Severity 3 1 (test)
5.42ECE
8
CIFAR-100-C Severity 1 (test)
3.35ECE
8
CIFAR-100 50% deletion of oak_tree class 52 (train test retain)
78.5Utility Accuracy (UA)
8
CIFAR-100 apple class, 50% deletion (forget)
97.1Utility Accuracy (UA)
8