Figure 6.

SVM performs better than NN for larger superfamily sizes. The figure plots the accuracy of NN and SVM classifiers (y-axis), averaged over all superfamilies less than or equal to a given size (x-axis) for classification from (A) sequence and (B) structure.