Endocrine-Immune Network and Its Application for Optimization - Intelligent Computing for Sustainable Energy and Environment - page 152

3.4 Antibody Learning

In the TSP, the process of antibody learning is performed in the following way:
Select a segment route from memory antibody Ab m randomly, and then add the
segment to the end of the learning antibody Ab i . Besides, remove the duplicate
city number of antibody Ab i . As shown in Fig.3.

Ab

i

2 5 7 8 3 1 6 4 3 6 8

learning

Ab

2 5 3 6 8 1 7 4

m

Fig. 3. antibody learning operator

3.5 Enzymatic Reaction

Here, we apply a greedy algorithm, the Interpolation-based Optimization Method
for Antibody algorithm, to implement the enzymatic reaction of endocrine sys-
tem for immune network, whose aim is to improve the convergence. According
to the characteristics of TSP, the interpolation-based optimization method is
designed as follows: Select the first visited city c j ( Ab i )in Ab i , and try to insert
it into next position. If it can shorten the route, the action will be performed; if
not, undo it. The operator is repeated until the route can't be shortened. The
pseudocode of the interpolation-based optimization method is given below.
————————————————————————————————
Algorithm: the Interpolation-based
Optimization Method for Antibody
do while Path can shorten
for n=1 to k do
for j=2 to k-1 do
if
[ d k, 1 ( Ab i )+ d 1 , 2 ( Ab i )+ d j,j +1 ( Ab i )] − [ d j, 1 ( Ab i )+ d 1 ,j +1 ( Ab i )+ d k, 2 ( Ab i )] > 0
then
remove c 1 ( Ab i )
insert c 1 ( Ab i ) between c j ( Ab i )and c j +1 ( Ab i )
break
end if
end for
if Path has not been modified then
move c 1 ( Ab i ) to the end of the Ab i
end if
end for
end do
—————————————————————————————————-
To summary, the whole pseudocode of the proposed algorithm EINET-TSP
is presented in the following.