Fig. 6
From: Detection of event-related potentials in individual subjects using support vector machines
Effects of electrode selection and temporal window size on a linear SVM trained with feature vectors that contain all three scalp electrodes (average of all available epochs). As in Fig. 5, increasing the window width systematically improved classifier accuracy. At the largest window size, the Frontal site (dotted line with diamond), Central site (dashed line with square), and the feature vector containing data from all sites (solid line with circle) produced significantly better performance than the Parietal site (dash-dotted line with triangle). Nearly equal performance was obtained for the Frontal site, Central site, and all sites. For clarity, select error bars are displayed around 200, 500, and 900-ms window sizes. Error bars indicate standard deviation