A Learning Automata-Based Multiobjective Hyper-Heuristic

Wenwen Li

doi:10.1109/TEVC.2017.2785346

Return

A Learning Automata-Based Multiobjective Hyper-Heuristic

2019-02-01

Wenwen Li *

Ender Özcan

Robert John

DOI：10.1109/TEVC.2017.2785346

Original

Save

View PDF

Abstract

En 中文

Metaheuristics, being tailored to each particular domain by experts, have been successfully applied to many computationally hard optimization problems. However, once implemented, their application to a new problem domain or a slight change in the problem description would often require additional expert intervention. There is a growing number of studies on reusable cross-domain search methodologies, such as selection hyper-heuristics, which are applicable to problem instances from various domains, requiring minimal expert intervention or even none. This paper introduces a new learning automata-based selection hyper-heuristic controlling a set of multiobjective metaheuristics. The approach operates above three well-known multiobjective evolutionary algorithms and mixes them, exploiting the strengths of each algorithm. The performance and behavior of two variants of the proposed selection hyper-heuristic, each utilizing a different initialization scheme are investigated across a range of unconstrained multiobjective mathematical benchmark functions from two different sets and the real-world problem of vehicle crash-worthiness. The empirical results illustrate the effectiveness of our approach for cross-domain search, regardless of the initialization scheme, on those problems when compared to each individual multiobjective algorithm. Moreover, both variants perform significantly better than some previously proposed selection hyper-heuristics for multiobjective optimization, thus significantly enhancing the opportunities for improved multiobjective optimization.

Keywords:

Evolutionary algorithms

hyper-heuristics

multiobjective optimization

online learning

operational research

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.