Evolutionary hyper-heuristics for improving global optimization algorithms

A. Manescu,Bogdan Dumitrescu
DOI: https://doi.org/10.1109/CSCS59211.2023.00009
2023-05-01
Abstract:In this paper, a new global optimization methodology is proposed. The approach presented is a hyper-heuristic, namely Evolutionary Bald Eagle Search (EBES), based on the Bald Eagle Search (BES) optimization algorithm. Additionally, for comparison, Evolutionary Particle Swarm optimization (EPSO) is demonstrated, which is very similar with EBES in the overall approach, but is based on the well-known PSO heuristic. The main methodology consists in a genetic algorithm that is employed as a high-level method to adjust the hyper-parameters of the low-level heuristics in an online manner in order to maximize the performance of the search for global optima. The proposed hyper-heuristics are compared with the heuristics on which they are based, showing a significant improvement in performance.
Computer Science,Engineering
What problem does this paper attempt to address?