Progressively Interactive Evolutionary Multi-Objective Optimization ...
Progressively Interactive Evolutionary Multi-Objective Optimization ...
Progressively Interactive Evolutionary Multi-Objective Optimization ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
DM Calls and Func. Evals. (in thousands)<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
Accuracy<br />
# of DM Calls<br />
Func. Evals./1000<br />
2 5 10<br />
Frequency of DM Calls<br />
20<br />
0.05<br />
0.04<br />
0.02<br />
0<br />
Accuracy<br />
Fig. 9. Three performance measures on modified<br />
ZDT1 problem for different τ values.<br />
DM Calls and Func. Evals. (in thousands)<br />
60<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
2<br />
Accuracy<br />
# of DM Calls<br />
Func. Evals./1000<br />
5 10<br />
Frequency of DM Calls<br />
20<br />
0.3<br />
0.2<br />
0.1<br />
0<br />
Accuracy<br />
Fig. 10. Three performance measures on threeobjective<br />
modified DTLZ2 problem for different<br />
τ values.<br />
preferred point, thereby causing a large number of generations to satisfy termination<br />
conditions and a large number of DM calls.<br />
Figure 10 shows the same three performance measures on the three-objective modified<br />
DTLZ2 problem. For this problem, the number of DM calls is minimum for τ = 5<br />
and accuracy and the number of function evaluations are also better for τ = 5 generations.<br />
Once again, too small or too large τ is found to be detrimental.<br />
Based on these simulation studies on two and three-objective optimization problems,<br />
one can conclude that a value of τ close to 5 generations is better in terms of<br />
an overall performance of the PI-NSGA-II procedure. This value of τ provides a good<br />
convergence accuracy, requires less function evaluations, and less DM calls to converge<br />
near the most preferred point.<br />
7 Conclusions<br />
In this paper, we have proposed a preference based evolutionary multi-objective optimization<br />
(PI-EMO) procedure which uses a polyhedral cone to modify domination. It<br />
accepts preference information from the decision maker in terms of the best solution<br />
from the archive set. The preference information from the decision maker and information<br />
from the non-dominated set of the parent population of the evolutionary algorithm<br />
have been used together to construct a polyhedral cone. Progressive information from<br />
the population of the evolutionary algorithm as well as the decision maker is used to<br />
modify the polyhedral cone after every few iterations. This approach helps in approaching<br />
towards the most preferred point on the Pareto-front by focussing the search on the<br />
region of interest.<br />
The direction provided by the cone has been used to develop a termination criterion<br />
for the algorithm. The procedure has then been applied to three different test-problems<br />
involving two, three and five objectives. The procedure has been successful in finding<br />
the most preferred solution corresponding to the DM-emulated utility function. A para-<br />
70