ADBench
Benchmarks
TaskDatasetSOTA resultTrendResults
ADBench noisy-context scenario 1.0
84.78F1 Score
532
ADBench Tabular (aggregated across 47 datasets)
86.025Average AUROC
41
ADBench
90.43Mean AUCROC
34
ADBench MNIST-C
21.86F1 Score
26
ADBench breastw
94.15F1 Score
26
ADBench aloi
10.36Average AUC PR
26
ADBench v1 (train)
0.147ALOI
19
ADBench 57 Datasets (test)
54.7AP
19
ADBench ID 2
92.68AUCROC
17
ADBench ID 1
66.95AUCROC
17
ADBench anomaly ratio categories
31.3AP (< 3%)
15
ADBench unsupervised (contaminated) setting
34.3AUC-PR
15
ADBench 38 real-world datasets (clean-context)
65.52F1 Score
14
ADBench All Datasets 57 datasets
81.49ROC-AUC
10
ADBench Low Dimensionality, d ≤ 20 (24 datasets)
90.14ROC-AUC
10
ADBench Medium Dimensionality (20 < d ≤ 100)
80.13ROC-AUC
10
ADBench High Dimensionality d > 100 (15 datasets)
73.98ROC-AUC
10
ADBench 100% labeled anomaly ratio
2.2Mean Rank (Rk)
9
ADBench 25% labeled anomaly ratio
2.5Mean Rank (Rk)
9
ADBench 5% labeled anomaly ratio
2.9Mean Rank
9
ADBench 1% labeled anomaly ratio
2.7Mean Rank
9
ADBench Novelty
2.6Mean Rank
9
ADBench One-Class
3.3Mean Rank
9
ADBench Unsupervised
3.9Mean Rank
9
ADBench annthyroid
82.88Average AUC PR
9