A electricity theft detection method through contrastive learning in smart grid

EURASIP Journal on Wireless Communications and Networking

Table 3 The comparison results with the base models

Models	Training ratio 50%		Training ratio 60%		Training ratio 70%		Training ratio 80%
Models	AUC	Recall	AUC	Recall	AUC	Recall	AUC	Recall
WDCNN [4]	0.6690	0.5889	0.6797	0.5905	0.6721	0.6359	0.6832	0.6339
HybridAttn [20]	0.7295	0.6096	0.7455	0.6286	0.7440	0.6055	0.7561	0.6284
GCN-CNN [9]	0.7810	–	0.7760	–	0.7870	–	0.787	–
HORLN [21]	0.7844	0.7236	0.8021	0.7379	0.8080	0.7583	0.8155	0.7515
Proposed method	0.7854	0.7590	0.7982	0.7922	0.8101	0.7991	0.8159	0.8136