Table 5. Performance comparison of different methods on the LEAD1.0 dataset (FPR = 0.5%).

Dataset	Model	Precision	Recall	F1-score	Accuracy
LEAD1.0	SVM (Depuru, Wang & Devabhaktuni, 2011)	0.900	0.562	0.692	0.778
	Wide & Deep CNN (Zheng et al., 2017)	0.928	0.800	0.859	0.883
	LSTM (Munawar et al., 2021)	0.930	0.825	0.874	0.894
	CNN-LSTM (Almazroi & Ayub, 2021)	0.933	0.875	0.903	0.917
	Autoencoder (Takiddin et al., 2022)	0.935	0.900	0.917	0.928
	Transformer (Ours)	0.938	0.950	0.944	0.950