Research Article

An Adaptive Calibration Framework for mVEP-Based Brain-Computer Interface

Table 4

Performance of both calibrations when the threshold for reliable sample selection takes different values.

SubjectsAdaptive calibration by SVM and fCM (accuracy (%)/ITR)SVM
0.60.650.70.750.8

S183.3/12.483.3/12.486.1/13.486.1/13.483.3/12.483.3/12.4
S294.4/17.191.7/15.894.4/17.194.4/17.194.4/17.191.7/15.8
S372.2/8.769.4/7.969.4/7.969.4/7.966.7/7.266.7/7.2
S494.4/17.191.7/15.891.7/15.894.4/17.194.4/17.191.7/15.8
S580.6/11.480.6/11.480.6/11.477.8/10.477.8/10.475/9.5
S688.9/14.688.9/14.691.7/15.891.7/15.891.7/15.888.9/14.6
S788.9/14.694.4/17.191.7/15.894.4/17.191.7/15.891.7/15.8
S891.7/15.891.7/15.894.4/17.194.4/17.191.7/15.888.9/14.6
S997.2/18.794.4/17.194.4/17.194.4/17.194.4/17.194.4/17.1
S1080.6/11.483.3/12.480.6/11.483.3/12.480.6/11.477.8/10.4
S1194.4/17.194.4/17.194.4/17.191.7/15.891.7/15.891.7/15.8

Mean ± std//////

denotes that the adaptive calibration method results are significantly higher than those of SVM approach (, paired -test).