3.9 shows the confusion matrix of a RBFNN model constructed

eds data [Ajaz and Hussain, 2015]. Figure 3.27 shows the ROC

this model.

9. The confusion matrix of the RBFNN model constructed for the seeds data.

A

B

%

A

61

10

85.9

B

8

57

87.7

%

88.4

85.1

85.7

.27. The ROC curve of the RBFNN model constructed for the seeds data.

e bio-basis function neural network algorithm

chine learning algorithms except for the decision tree algorithms

to model numerical data. Therefore, how to deal with non-

l variables such as the amino acids of a sequence data set requires

experimental design.

opularly used approach is the binary encoding of the amino acids

cleic acids in peptides [Wu, et al., 1995; Kawabata and Doi, 1997;

nd Seffens, 1998]. With this approach, each nucleic acid is

by a four-bit long binary vector and each amino acid is encoded

it long binary vector. For instance, the amino acid alanine (A) is

d by 0000000000 0000000001 and the amino acid cysteine (C) is