MuEvo: LLM-Driven Evolution of Multi-Heuristic Ensemble

2026-08-04

AI Summary

This work addresses the limitation of existing large language model (LLM)-driven heuristic design approaches, which typically focus on optimizing individual components in isolation and struggle to coordinate interdependent heuristics within practical optimization frameworks. To overcome this, the authors propose MuEvo, a novel framework that jointly optimizes multiple heuristics through dynamic component management with reversible lifecycles and a relation-guided co-evolution mechanism, guided by ensemble-level feedback. MuEvo integrates short-budget probing, multi-ensemble evaluation, cross-component information sharing, and adaptive budget allocation, leveraging an LLM to orchestrate the evolution of component populations, thereby effectively capturing interdependencies and preserving components with latent long-term potential. Experimental results demonstrate that MuEvo significantly outperforms both manually designed frameworks and current LLM-based automated heuristic design methods across four combinatorial optimization problems, showing broad applicability to controller-driven heuristic pools and functionally differentiated algorithmic components.

Abstract

Large language model-based automated heuristic design (LLM-AHD) has shown strong potential in discovering effective heuristics for combinatorial optimization problems. However, existing methods primarily optimize a single heuristic, whereas practical optimization frameworks often rely on multiple interacting components. Directly extending single-heuristic methods is challenging because early component selection can overlook components with late potential, while independent evolution ignores inter-component dependencies. We propose MuEvo, an LLM-driven framework for evolving heuristic ensembles under ensemble-level feedback. MuEvo combines Dynamic Component Management, which uses short-budget probing and a reversible lifecycle to revise component priorities throughout the search, with LLM-Driven Co-Evolution, which coordinates component populations through Multi-Ensemble Evaluation, Cross-Component Information Sharing, Relation-Guided Pair Evolution, and Adaptive Budget Allocation. We evaluate MuEvo on selection hyper-heuristics and componentized ant colony optimization across four combinatorial optimization domains. Results show that MuEvo consistently improves human-designed frameworks and outperforms representative multi-component extensions of state-of-the-art LLM-AHD methods, demonstrating its effectiveness across both controller-mediated heuristic pools and functionally differentiated algorithmic components.

Problem

• combinatorial optimization
• heuristic ensemble
• LLM-driven evolution
• multi-component optimization
• inter-component dependencies

Innovation

• LLM-driven co-evolution
• dynamic component management
• multi-heuristic ensemble
• combinatorial optimization
• adaptive budget allocation