Table 10 Performance comparison of HMOA-GNN and existing CCFD methods on Simulated Credit Card Transactions dataset.

From: HMOA-GNN: adaptive adversarial GraphSAGE with hierarchical hybrid sampling and metric-optimized graph construction for credit card fraud detection

Ref

Algorithm

Accuracy

Precision

Recall

F1-score

AUC

Zou et al.49

DAE

0.8790

0.1809

0.8226

0.2965

0.9293

Pang et al.50

DevNet

0.8116

0.1339

0.9274

0.2340

0.9210

Ileberi et al.51

GA-RF

0.9815

0.7551

0.5968

0.6667

0.9504

Ni et al.27

SOBT

0.9800

0.9348

0.4300

0.5890

0.9760

Xiang et al.52

GTAN

0.8838

0.6581

0.4063

0.5535

0.8379

Chang et al.53

TFD

0.9807

0.8889

0.4800

0.6234

0.9672

Ours

HMOA-GNN

0.9815

0.7049

0.6935

0.6992

0.9177

  1. Bold values indicate the best performance.
Back to article page