The distribution of the target values of a simulated data set using GP to solve a

proximation problem.

the 13^th cycle, the stabilisation criterion was satisfied and the

process was terminated for this function approximation problem.

elite RPN chromosome was zxെzy+*, which was exactly what

cted. Figure 8.22(a) shows how the sum of the squared errors

with the learning cycle and Figure 8.22(b) shows how the fitness

ment evolved with the learning cycle.

(a) (b)

a) The sum of the squared errors evolved with the learning cycle. (b) The fitness

nts evolved with the learning cycle.

e 8.23 shows how the predicted values were correlated with the

ues based on the top model in the elite set. It can be seen that two

lues have an extremely high correlation.