"... In PAGE 2: ... The predictive performance of the resulting system is shown by the fact that, if PFAM-SEED 8.0 is used as the reference database, over 60% of the groups had none or only one mistaken prediction ( Table1 ), and the average difference in the domain boundaries is lt;5 amino acids in 90% of the cases (Figure 2). IMPROVEMENTS WITH RESPECT TO THE PREVIOUS RELEASE (i) The examples included in the consolidated domain sequence collection SBASE A were filtered so as to discard conspicuously short or long domain examples.... ..."

"... In PAGE 4: ... The building tree method introduces a bias, which we intuitively describe below. Let us consider the learning set given in Table1 . It can be classi ed using the four production rules in Fig.... In PAGE 6: ...E, which corresponds to the set of parts of the training set. Let us, for instance, consider the small training set given in Table1 . ((A = a1) ^ (B = b3)) belongs to D, whereas fE1 E2 E3g belongs to E.... In PAGE 7: ... Once a training example is a ected to a leaf (or to a subtree), it cannot be found in any other leaf (or subtree). Let us consider the learning set in Table1 . Given the decision tree in Fig.... ..."

"... In PAGE 16: ... The learning parameters are shown in Table 5. Results from these experiments can be seen in Table6 , Table 7 and Table 8 with their graphical representation in Figure 1, Figure 2 and Figure 3 respectively Tables 9, 10, 11 and 12 show stable behaviour with respect to most learning parameters, although each parameter has an interval that gives optimal results. Table 9: Sensitivity of LVQ1 with respect to learning rate.... ..."

"... In PAGE 5: ... 5 Example A small arti cial learning domain is here included to show the presented con- cepts. In Table1 the learning set is presented. There are 10 examples with 3 attributes of the quality type.... ..."