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

Table 3

Classification accuracy using NN among all subjects.

Feature combination	S1	S2	S3	S4	S5	S6	S7
Mean & slope	94.4	94.6	95.5	95.1	94.4	93.5	94.6
Mean & peak	95.9	96.2	97.2	97.6	97.6	97.1	96.7
Mean & variance	89.1	92.6	91.5	91.3	90.7	89.6	94.0
Slope & peak	92.5	91.1	91.7	95.9	93.1	93.6	92.6
Slope & variance	95.0	95.1	94.0	93.7	93.2	92.5	93.6
Peak & variance	88.2	88.7	86.8	93.5	88.6	90.2	86.3
Peak & skewness	64.4	66.9	64.5	65.4	62.5	58.0	61.0
Mean & skewness	53.7	57.7	55.6	54.1	53.7	52.2	58.6
Slope & skewness	50.8	47.8	51.4	50.1	49.9	51.3	53.8
Kurtosis & skewness	47.3	50.7	55.0	51.3	54.3	60.1	54.8
Variance & skewness	50.6	48.1	51.2	49.7	49.3	51.4	54.7
Peak & kurtosis	65.4	63.4	59.2	65.7	62.7	55.0	60.7
Mean & kurtosis	53.5	55.0	52.9	52.1	52.2	51.3	53.7
Slope & kurtosis	52.1	49.8	50.7	48.6	50.9	48.8	50.7
Variance & kurtosis	51.9	50.4	50.6	48.9	50.6	48.8	50.7