Skip to main content
. 2016 Nov 21;7:197. doi: 10.3389/fneur.2016.00197

Figure 4.

Figure 4

The classification performance as a function of number of channels selected via the FCSI, SFS, and FCSI + SFS methods, respectively. For each subject, the classification accuracies were derived from the testing dataset. The classification accuracies from all 12 subjects were averaged and plotted with SD error bars. The classification accuracies derived from applying both WPT and TD features to 10 predefined channels are indicated as two horizontal dashed lines.