ReVEL: Multi-Turn Reflective LLM-Guided Heuristic Evolution via Structured Performance Feedback
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910259863355392 |
|---|---|
| author | Van Duc, Cuong Tuan, Minh Nguyen Dinh Duc, Tam Vu Duy, Tung Vu Van, Son Nguyen Thi, Hanh Nguyen Thanh, Binh Huynh Thi |
| author_facet | Van Duc, Cuong Tuan, Minh Nguyen Dinh Duc, Tam Vu Duy, Tung Vu Van, Son Nguyen Thi, Hanh Nguyen Thanh, Binh Huynh Thi |
| contents | Designing effective heuristics for NP-hard combinatorial optimization problems remains challenging and often requires substantial domain expertise. Recent LLM-guided evolutionary methods have shown promise for automated heuristic generation, but most existing approaches refine heuristics independently or through limited pairwise feedback. We propose ReVEL: Multi-Turn Reflective LLM-Guided Heuristic Evolution via Structured Performance Feedback, a framework for group-wise multi-turn heuristic refinement. ReVEL organizes heuristics into behavior-aware reflective groups, including similarity-driven groups for localized refinement and diversity-driven groups for exploratory search. Within each group, the LLM performs iterative multi-turn refinement using accumulated performance feedback, enabling related heuristics to be jointly analyzed and progressively improved across evolutionary iterations. Experiments on standard combinatorial optimization benchmarks show that ReVEL generally improves optimization performance over existing LLM-guided evolutionary baselines across multiple settings and LLM backbones. Additional analyses suggest that behavior-aware grouping contributes to more consistent refinement trajectories during iterative heuristic evolution. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_04940 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ReVEL: Multi-Turn Reflective LLM-Guided Heuristic Evolution via Structured Performance Feedback Van Duc, Cuong Tuan, Minh Nguyen Dinh Duc, Tam Vu Duy, Tung Vu Van, Son Nguyen Thi, Hanh Nguyen Thanh, Binh Huynh Thi Artificial Intelligence Designing effective heuristics for NP-hard combinatorial optimization problems remains challenging and often requires substantial domain expertise. Recent LLM-guided evolutionary methods have shown promise for automated heuristic generation, but most existing approaches refine heuristics independently or through limited pairwise feedback. We propose ReVEL: Multi-Turn Reflective LLM-Guided Heuristic Evolution via Structured Performance Feedback, a framework for group-wise multi-turn heuristic refinement. ReVEL organizes heuristics into behavior-aware reflective groups, including similarity-driven groups for localized refinement and diversity-driven groups for exploratory search. Within each group, the LLM performs iterative multi-turn refinement using accumulated performance feedback, enabling related heuristics to be jointly analyzed and progressively improved across evolutionary iterations. Experiments on standard combinatorial optimization benchmarks show that ReVEL generally improves optimization performance over existing LLM-guided evolutionary baselines across multiple settings and LLM backbones. Additional analyses suggest that behavior-aware grouping contributes to more consistent refinement trajectories during iterative heuristic evolution. |
| title | ReVEL: Multi-Turn Reflective LLM-Guided Heuristic Evolution via Structured Performance Feedback |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2604.04940 |