TABLE 1.

The classification accuracies on MI and MA tasks obtained by fusing EEG and fNIRS multimodal features based on the proposed fusion method and the direct splicing method for 29 subjects.

Task	MI		MA
Features Subject	EEG-fNIRS (Direct splicing)	EEG-fNIRS (Proposed fusion method)	EEG-fNIRS (Direct splicing)	EEG-fNIRS (Proposed fusion method)
1	81.49%	95.03%	91.26%	100.00%
2	80.69%	97.11%	85.78%	99.54%
3	79.69%	96.99%	77.97%	94.94%
4	83.59%	97.21%	83.03%	98.31%
5	80.79%	96.18%	81.54%	96.51%
6	74.25%	97.73%	86.62%	99.48%
7	84.63%	98.23%	88.64%	97.76%
8	71.39%	96.35%	82.69%	98.32%
9	82.93%	99.01%	86.78%	98.39%
10	86.13%	98.89%	98.63%	98.31%
11	77.23%	97.13%	83.52%	97.61%
12	75.90%	97.37%	79.96%	93.47%
13	76.45%	98.23%	79.44%	96.74%
14	77.97%	98.36%	78.07%	99.68%
15	85.30%	95.08%	83.16%	97.12%
16	74.96%	96.37%	79.17%	99.77%
17	83.60%	97.93%	86.56%	98.88%
18	77.33%	96.44%	91.97%	99.80%
19	81.49%	98.73%	84.56%	98.84%
20	74.51%	98.33%	80.45%	99.23%
21	89.53%	94.10%	84.84%	98.97%
22	85.54%	95.64%	87.88%	98.58%
23	84.81%	98.34%	87.64%	99.73%
24	77.19%	94.51%	87.42%	99.22%
25	82.85%	90.13%	89.76%	99.85%
26	83.65%	97.42%	90.27%	98.54%
27	83.87%	97.24%	87.90%	99.21%
28	71.46%	98.06%	81.72%	97.68%
29	84.48%	93.18%	90.14%	99.75%
Mean	80.47%	96.74%	85.43%	98.42%
Standard deviation	4.68%	1.96%	4.79%	1.52%