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

Table 2

Classification accuracy using QDA among all subjects.

Feature combination	S1	S2	S3	S4	S5	S6	S7
Mean & slope	95.5	96.5	95.5	96.6	96.0	95.4	96.9
Mean & peak	97.0	98.4	97.7	98.2	97.4	98.4	98.4
Mean & variance	97.4	98.0	96.0	96.5	96.1	95.4	97.2
Slope & peak	94.4	95.0	88.5	95.7	95.1	93.6	96.1
Slope & variance	98.0	98.2	94.0	95.5	94.4	93.0	95.0
Peak & variance	97.0	98.1	93.7	97.4	96.0	94.1	96.9
Peak & skewness	90.6	89.8	83.8	92.6	89.7	88.8	86.8
Mean & skewness	91.0	91.5	90.5	91.8	89.1	87.6	93.2
Slope & skewness	89.3	89.7	84.2	88.0	90.0	86.6	91.2
Kurtosis & skewness	48.3	53.7	51.4	52.8	50.2	56.1	52.1
Variance & skewness	97.5	97.9	90.7	95.6	94.7	90.7	92.1
Peak & kurtosis	89.6	88.2	82.7	92.1	88.1	88.2	86.2
Mean & kurtosis	89.2	91.1	88.8	90.8	88.8	87.5	93.9
Slope & kurtosis	89.5	89.6	84.3	87.7	89.8	86.6	91.0
Variance & kurtosis	97.6	97.9	90.6	95.4	94.4	90.6	91.7