Progressively Interactive Evolutionary Multi-Objective Optimization ...
Progressively Interactive Evolutionary Multi-Objective Optimization ...
Progressively Interactive Evolutionary Multi-Objective Optimization ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
3. Thepreviousapproacheshadafixedpopulationsizeforthelowerlevel. Thismade<br />
the number of upper level variable vectors fixed from one generation to another<br />
for these approaches, whereas in H-BLEMO, the use of a variable population size<br />
for different lower level NSGA-IIs establishes that the upper level population can<br />
hold varying number of upper level variable vectors. This allows the H-BLEMO<br />
proceduretoprovideavaryingimportancebetweentheextentofupperandlower<br />
level optimizations, as maybe demandedby aproblem.<br />
4. The earlierproceduresdid not have aalgorithmically sensible terminationcriteria<br />
fortheupperorthelowerlevelNSGA-IIandhadtorunforafixednumberofgenerations.<br />
In H-BLEMO a hypervolume based termination criteria has been used<br />
where both lower and upper level terminate, based on the dynamic performance<br />
of the algorithms. The criteriaensures the termination of eachNSGA-IIwhenever<br />
the corresponding non-dominated front has stabilized,therebyavoiding unnecessaryfunction<br />
evaluations.<br />
6 Results onTest Problems<br />
We use the following standard NSGA-II parameter values in both lower and upper<br />
levelsonallproblemsof this study: Crossoverprobabilityof0.9,distributionindex for<br />
SBX of 15, mutation probability of 0.1, distribution index for the polynomial mutation<br />
of 20. The upper level population size is set proportional to the total number of variables<br />
(n): Nu = 20n. As described in the algorithm, all other parameters including the<br />
lower level population size, termination criteria are all set in a self-adaptive manner<br />
during the optimization run. In all cases, we have used 21 different simulations startingfromdifferentinitialpopulationsandshowthe0,50,and100%attainmentsurfaces<br />
(Fonseca andFleming, 1996)todescribethe robustness of the proposedprocedure.<br />
6.1 ProblemTP1<br />
This problem has three variables (one for upper level and two for lower level). Thus,<br />
we use Nu = 60 population members. Figure 11 shows the final archive members of<br />
a single run on the upper level objective space. It can be observed that our proposed<br />
procedure is able to find solutions on the true Pareto-optimal front. Conditions for<br />
the exact Pareto-optimal solutions are given in equation (3). For each of the obtained<br />
solutions, we use the upper level variable (y) value to compute lower level optimal<br />
variable (x ∗ 1 and x∗ 2<br />
) values using the exact conditions for Pareto-optimality and com-<br />
pare the values with H-BLEMO solutions. The average error 2<br />
i=1 (xi − x ∗ i )2 /2 for<br />
eachobtained solution is computed and then averagedover all archivesolutions. This<br />
error value for our H-BLEMO is found to be 7.0318 × 10 −5 , whereas the same error<br />
value computed for the solutions reported with our earlier algorithm (Deb and Sinha,<br />
2009a) is found to be 2.2180 × 10 −3 . The closer adherence to optimal variable values<br />
withH-BLEMOindicatesabetterperformanceofourhybridalgorithm. Theminimum,<br />
medianand worst number of function evaluations neededover 21 differentruns of H-<br />
BLEMO are 567,848, 643,753, and 716,780, respectively. Although for a three-variable<br />
problem,thesenumbers mayseemtoolarge,one needstorealizethatthebilevelproblems<br />
have one optimization algorithm nested into another and Pareto-optimality for<br />
the lower level problem is essential for an upper level solution. In comparison, our<br />
previous algorithm (Sinha and Deb, 2009) required 1,030,316, 1,047,769 and 1,065,935<br />
functionevaluations forbest,medianandworst performingruns. Despite using about<br />
40%less functionevaluations, our H-BLEMOalsofinds more accuratesolutions.<br />
97