Table 2.

The comparison summary of classification performance in proposed models.

Datasets	Methods	Accuracy (%)	Kappa	F1 Score
BCI-IV2a	FBCSP [17]	67.80	NA *	0.675
	ShallowConvNet [29]	72.92	0.639	0.728
	DeepConvNet [11]	70.10	NA	0.706
	EEGNet [20]	72.40	0.630	NA
	CP-MixedNet [26]	74.60	NA	0.743
	TS-SEFFNet [29]	74.71	0.663	0.757
	MBEEGNet [37]	82.01	0.760	0.822
	MBShallowCovNet [37]	81.15	0.749	0.814
	CNN + BiLSTM (fixed) [15]	75.81	NA	NA
	Proposed (MBEEGSE)	82.87	0.772	0.829
HGD	FBCSP [17]	90.90	NA	0.914
	ShallowConvNet [29]	88.69	0.849	0.887
	DeepConvNet [11]	91.40	NA	0.925
	EEGNet [37]	93.47	0.921	0.935
	CP-MixedNet [26]	93.70	NA	0.937
	TS-SEFFNet [29]	93.25	0.910	0.901
	MBEEGNet [37]	95.30	0.937	0.954
	MBShallowCovNet [37]	95.11	0.935	0.951
	CNN + BiLSTM (fixed) [15]	96.00	NA	NA
	Proposed (MBEEGSE)	96.15	0.949	0.962

* NA means Not Available.