213783.fig.003a
(a)
213783.fig.003b
(b)
Figure 3: All potential hyperplanes are evaluated during the training phase of SVM, and the hyperplane chosen is the one which maximizes the separation margin (a). Once the algorithm is trained based on data with known labels, the hyperplane yields a classification rule which is capable of discriminating samples with unknown labels (training dataset). Unknown data points are represented by gray circles (b).