Large language model-driven evolutionary optimization: A hierarchical interaction framework

doi:10.1016/j.eswa.2026.134167

Large language model-driven evolutionary optimization: A hierarchical interaction framework

2026-08-27

0

PRE

AI

DOI：10.1016/j.eswa.2026.134167

Original

Original request for help

Save

Abstract

En 中文

• Proposes LLDEOF for controlled intermittent interaction between LLMs and EAs. • Designs semantic and relative-delta perturbations for discrete and continuous search. • Uses validation, repair, fallback, and greedy acceptance to improve reliability. • Evaluates 0-1 KP, MKP, and CEC2020 with paired statistical analysis. • Reports solution quality, LLM overhead, ablation, and parameter sensitivity.

Keywords:

Large language models

Evolutionary optimization

LLM-assisted intervention

Particle swarm optimization

Constrained optimization

Knapsack problem

Large language model-driven evolutionary optimization: A hierarchical interaction framework

Abstract

Journal

Organization