TY - GEN
T1 - Tailoring ε-MOEA to concept-based problems
AU - Moshaiov, Amiram
AU - Snir, Yafit
PY - 2012
Y1 - 2012
N2 - Concept-based MOEAs are tailored MOEAs that aim at solving problems with a-priori defined subsets of solutions that represent conceptual solutions. In general, the concepts' subsets may be associated with different search spaces and the related mapping into a mutual objective space could have different characteristics from one concept to the other. Of a particular interest are characteristics that may cause premature convergence due to local Pareto-optimal sets within at least one of the concept subsets. First, the known ε-MOEA is tailored to cope with the aforementioned problem. Next, the performance of the new algorithm is compared with C 1-NSGA-II. Concept-based test cases are devised and studied. In addition to demonstrating the significance of premature convergence in concept-based problems, the presented comparison suggests that the proposed tailored MOEA should be preferred over C 1-NSGA-II. Suggestions for future work are also included.
AB - Concept-based MOEAs are tailored MOEAs that aim at solving problems with a-priori defined subsets of solutions that represent conceptual solutions. In general, the concepts' subsets may be associated with different search spaces and the related mapping into a mutual objective space could have different characteristics from one concept to the other. Of a particular interest are characteristics that may cause premature convergence due to local Pareto-optimal sets within at least one of the concept subsets. First, the known ε-MOEA is tailored to cope with the aforementioned problem. Next, the performance of the new algorithm is compared with C 1-NSGA-II. Concept-based test cases are devised and studied. In addition to demonstrating the significance of premature convergence in concept-based problems, the presented comparison suggests that the proposed tailored MOEA should be preferred over C 1-NSGA-II. Suggestions for future work are also included.
UR - http://www.scopus.com/inward/record.url?scp=84866408123&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-32964-7_13
DO - 10.1007/978-3-642-32964-7_13
M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???
AN - SCOPUS:84866408123
SN - 9783642329630
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 122
EP - 131
BT - Parallel Problem Solving from Nature, PPSN XII - 12th International Conference, Proceedings
T2 - 12th International Conference on Parallel Problem Solving from Nature, PPSN 2012
Y2 - 1 September 2012 through 5 September 2012
ER -