TABLE 1.

Hyperparameter sensitivity analysis results.

Experiment number	Learing rate	Weight decay	Batch size	Validation loss	Acc(%)	Recall (%)	F1 (%)
Experiment 1	1e-2	1e-3	16	0.3735	82.96	82.74	83.03
Experiment 2	1e-3	1e-3	16	0.0362	98.54	98.41	98.66
Experiment 3	1e-2	1e-3	32	0.2003	91.78	91.82	91.72
Experiment 4	1e-3	1e-3	32	0.0415	98.42	98.35	98.40
Experiment 5	1e-2	1e-4	16	0.3628	83.24	83.19	82.98
Experiment 6	1e-3	1e-4	16	0.0344	98.76	98.77	98.54
Experiment 7	1e-2	1e-4	32	0.0491	98.27	97.83	98.14
Experiment 8	1e-3	1e-4	32	0.0293	98.94	99.01	98.71