A Guidance Method for Robustness Surrogate Assisted Multi-objective Evolutionary Algorithms

Abstract

In the real world, multi-objective problems
(MOPs) are relatively common in optimization in the areas
of design, planning, decision support, etc. In fact, problems
include two or many objectives, there is a class of problems
called expensive problems that are problems with complex
mathematical models, large computational costs, etc. They
can not be solved by normal techniques, they are usually to
be solved with techniques such as simulation, decomposing,
problem transformation. In particular, using a surrogate
model with Kriging, neural networks techniques in combination with an evolutionary algorithm is a subtle choice,
with many positive results, being studied and applied in
practice. However, the use of a surrogate model with Kriging,
neural networks combining selection strategy, sampling... can
reduce the robustness of the algorithms during the search.
This paper analyzes the issues affecting the robustness of
the multi-objective evolutionary algorithms (MOEAs) using
surrogate models and suggests the use of a guidance technique to increase the robustness of the algorithm, through
analysis, experiment and results are competitive and effective
to improve the quality of MOEAs using a surrogate model to
solve expensive problems.