Table 5 The balanced accuracy results (%) of the proposed model with our sampling method, against the SMOTE sampling method over all datasets using the first scenario
From: Deep reinforcement learning for data-efficient weakly supervised business process anomaly detection
Datasets | Our sampling method | SMOTE |
|---|---|---|
DS1 | 98.90 ± 0.07 | 93.06 ± 0.06 |
DS2 | 97.02 ± 0.05 | 90.48 ± 0.07 |
BPIC12 | 93.37 ± 0.06 | 84.53 ± 0.08 |
BPIC17 | 95.04 ± 0.04 | 88.91 ± 0.12 |