Effective Dimension Extraction Mechanism: A novel mechanism for meta-heuristic algorithms in solving complex high-dimensional problems
•Proposing a mechanism to improve heuristics by extracting effective dimensions.•Using manifold learning to filter redundant dimensions in particles.•Mapping particles to high-dimensional space while preserving distance relations.•Based on the two mappings, Effective Dimension Extraction Mechanism i...
Uloženo v:
| Vydáno v: | Expert systems with applications Ročník 285; s. 127733 |
|---|---|
| Hlavní autoři: | , , |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
Elsevier Ltd
01.08.2025
|
| Témata: | |
| ISSN: | 0957-4174 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | •Proposing a mechanism to improve heuristics by extracting effective dimensions.•Using manifold learning to filter redundant dimensions in particles.•Mapping particles to high-dimensional space while preserving distance relations.•Based on the two mappings, Effective Dimension Extraction Mechanism is designed.•EDEM enhances DE and PSO, demonstrating its value in meta-heuristic optimization.
In recent years, meta-heuristic optimization algorithms have been increasingly studied due to their wide applicability in practical applications. However, the interference of redundant dimensions in complex high-dimensional problems often leads to performance degradation. To solve this problem, a new Effective Dimension Extraction Mechanism (EDEM) is proposed, which extracts effective dimensional information during the optimization using a Manifold Learning-based Mapping from high-dimensional to low-dimensional space and a Feature Mapping from low-dimensional to the original high-dimensional space, thereby providing direct guidance information for meta-heuristic optimization algorithms to correct the evolution direction of particle populations. As an independent mechanism, EDEM can be easily applied to multiple meta-heuristic optimization algorithms, providing a new way for these algorithms to solve the problems. A series of experiments are conducted to validate the effectiveness and advancement of EDEM. First, experiments on fifteen complex high-dimensional problems demonstrate its effectiveness in improving DE, PSO and ASO performance. Next, the comparative evaluation experiments with the state-of-art PSO and DE variant algorithms are performed, and the results confirm that EDEM is an effective mechanism, which provides a promising solution for the optimization on complex high-dimensional problems. Finally, EDEM is applied to the Capacitated Vehicle Routing Problem, and the experimental results show that the improved meta-heuristic optimization algorithm of EDEM can solve this practical problem well, which shows the practical ability of this new mechanism. |
|---|---|
| ISSN: | 0957-4174 |
| DOI: | 10.1016/j.eswa.2025.127733 |