g

(

g

p

These predicted class labels are shown in the first two columns of

. In addition to these two LDAs, the third LDA is added on top

wo LDAs (namely, ߙ and ߚ). The third LDA examines whether

cted class labels of LDA ߙ are the same as the predicted class

LDA ߚ. If the predicted class label of LDA ߙ is the same as the

class label of LDA ߚ, the top level LDA outputs a predicted class

. Otherwise, the top level LDA outputs a predicted class label

terwards, four data points have been well-labeled or well-

. Table 3.7 shows this result. The process is shown in Figure

nd Table 3.7.

(a) (b)

An analysis of XOR problem using multiple LDAs leading to a MLP model.

R problem with two LDAs. (b) The connection of three LDA results in a MLP

Table 3.7. The analysis of three LDAs applied to the XOR problem.

LDA ߙ

LDA ߚ

LDA ߙ = ߚ

Output

a

0

1

No

0

b

1

0

No

0

c

1

1

Yes

1

d

1

1

Yes

1

he R packages for MLP

several R packages for constructing a MLP model. Typical ones

net, elmNN and brnn. These packages were developed based