Table 3.

Comparison of the subject-independent EEG decoding accuracy (%) with the present advanced classification approaches on the BCI competition IV 2a dataset and A1–A9 denotes nine different subjects.

Comparision method	Test subject (the remaining subjects used as training)									Mean	Std
	A1	A2	A3	A4	A5	A6	A7	A8	A9
EEGNet	53.76	39.54	54.88	43.02	51.80	48.96	60.70	61.38	47.82	51.32	6.94
CTCNN	55.90	26.04	70.66	45.49	33.16	35.42	40.97	61.29	60.07	47.67	14.20
AE XGboost	32.12	32.34	32.29	32.99	33.85	32.47	39.06	30.90	32.64	33.18	2.20
FBCSP	47.92	24.83	39.24	39.93	27.26	31.60	27.08	46.70	36.68	35.69	8.04
CRAM	61.02	42.35	73.11	50.43	50.74	51.48	67.26	69.72	66.85	59.22	10.13
CRNN-DF	65.51	45.18	78.62	53.58	55.64	56.03	71.28	75.02	70.78	63.52	10.70