Table 2.

Performance results of improved models.

Model	Training Time (h)	Test Time (s)	Single Test Time (s)	Test Accuracy (%)
CNN	2.0	0.25	0.0303	95.43
CNN-Res	4.33/1.43 (ES)	0.64	0.0343	98.24
LSTM	13.3	2.12	0.0115	97.10
LSTM-Res	18.7/8.6 (ES)	3.22	0.0312	99.67
GRU	13.8	2.45	0.0120	95.91
GRU-Res	18.9/11.7 (ES)	2.82	0.0318	99.49
Transformer	12.5	1.47	0.0126	71.68
Transformer-CNN	14.1	2.45	0.0149	98.96