Analysis of Different Classification Techniques for Two-Class Functional Near-Infrared Spectroscopy-Based Brain-Computer Interface

Table 6

Classification accuracy using ANN among all subjects.

Feature combination	S1	S2	S3	S4	S5	S6	S7
Mean & slope	95.6	97.4	94.9	97.1	96.6	96.0	95.6
Mean & peak	96.9	98.5	98.1	98.7	97.9	98.4	98.4
Mean & variance	98.0	98.4	98.1	97.7	98.2	96.4	98.2
Slope & peak	95.4	92.3	95.9	96.1	95.5	90.2	92.7
Slope & variance	97.9	98.4	97.7	98.2	98.0	97.0	98.0
Peak & variance	98.1	97.9	97.4	97.2	98.0	97.4	98.4
Peak & skewness	92.8	89.0	84.2	94.1	91.6	88.5	84.7
Mean & skewness	92.0	94.0	90.6	92.8	89.8	88.7	92.7
Slope & skewness	90.8	88.0	88.2	90.7	90.0	91.6	90.6
Kurtosis & skewness	54.5	55.7	53.6	55.6	56.2	55.7	51.6
Variance & skewness	98.4	98.5	97.6	98.4	97.2	97.0	98.0
Peak & kurtosis	90.1	88.5	81.4	94.1	87.8	87.5	84.9
Mean & kurtosis	90.1	93.5	89.0	91.6	90.7	89.8	91.5
Slope & kurtosis	91.3	88.1	88.2	90.0	90.6	91.9	90.3
Variance & kurtosis	98.4	98.4	97.1	97.4	97.6	96.6	95.5