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

Table 1

Classification accuracy using LDA among all subjects.

Feature combination	S1	S2	S3	S4	S5	S6	S7
Mean & slope	53.2	49.2	50.1	58.3	59.8	55.8	59.6
Mean & peak	94.5	96.7	90.3	92.0	91.1	92.2	94.9
Mean & variance	86.8	87.6	81.7	82.9	82.8	76.4	83.4
Slope & peak	87.3	83.6	80.8	85.9	83.8	83.6	81.2
Slope & variance	87.5	88.3	83.2	82.6	81.5	76.4	79.9
Peak & variance	89.7	89.8	83.7	87.5	87.3	83.7	81.2
Peak & skewness	89.1	83.6	80.4	86.5	81.6	83.2	81.2
Mean & skewness	49.6	50.6	48.6	53.3	53.6	53.5	50.1
Slope & skewness	50.5	51.2	50.3	53.8	54.0	53.1	50.9
Kurtosis & skewness	47.7	51.2	50.6	53.2	50.4	53.8	51.6
Variance & skewness	88.0	89.0	82.3	83.4	81.4	78.2	84.4
Peak & kurtosis	86.8	82.4	80.9	85.9	83.9	82.6	81.2
Mean & kurtosis	47.4	49.7	52.2	54.8	50.4	52.1	48.6
Slope & kurtosis	45.7	46.2	54.3	54.6	52.1	50.1	47.7
Variance & kurtosis	87.6	88.5	82.1	83.2	82.4	82.2	86.3