Improved ant colony optimization algorithms for continuous ... - CoDE
Improved ant colony optimization algorithms for continuous ... - CoDE
Improved ant colony optimization algorithms for continuous ... - CoDE
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
4.2 ACOMV Heuristics <strong>for</strong> Mixed-Variable Optimization Problems 35<br />
probability<br />
0.00 0.04 0.08 0.12<br />
f c d e b<br />
1 2 3 4 5 6 7 8 9<br />
ranks<br />
probability<br />
0.00 0.10 0.20 0.30<br />
Final Probabilities<br />
a b c d e f g<br />
categories<br />
Figure 4.2: Calculating probabilities of choosing different categorical values<br />
<strong>for</strong> a given decision variable. First, the initial probabilities are generated<br />
using a normal distribution and based on the best ranked solution that uses<br />
given value (left plot, dashed bars). Then, they are divided by the number<br />
of solutions using this value (left plot, solid bars), and finally a fixed value<br />
is added (right plot, dotted bars) in order to increase the probability of<br />
choosing those values, which are currently not used. The final probabilities<br />
are presented on the right plot, as solid bars.<br />
Di = {v i 1 , ..., vi c i}. The probability of choosing the l-th value is given by:<br />
o i l =<br />
wl<br />
cr=1 , (4.1)<br />
wr<br />
where wl is the weight associated with the l-th available value. It is calculated<br />
based on the weights ω and some additional parameters:<br />
wl = ωjl<br />
u i l<br />
+ q<br />
. (4.2)<br />
η<br />
The final weight wl is hence a sum of two components. The weight ωjl is<br />
calculated according to Equation 2.2, where the jl is the index of the highest<br />
quality solution that uses value vi l <strong>for</strong> the i-th categorical variable. In turn,<br />
ui l is the number of solutions using value vi l <strong>for</strong> the i-th categorical variable<br />
is, the lower is its<br />
in the archive. There<strong>for</strong>e, the more popular the value vi l<br />
final weight.<br />
The second component is a fixed value (i.e., it does not depend on the<br />
value vi l chosen): η is the number of values from the ci available ones that<br />
are unused by the solutions in the archive, and q is the same parameter of<br />
the algorithm that was used in Equation 2.2.