Beyond Average Performance: Dynamic Instance Clustering and Specialized Algorithm Design in LLM-Assisted Evolutionary Search

2026-08-04

AI Summary

This work addresses the limited robustness of existing large language model (LLM)-assisted evolutionary search methods, which overly prioritize average performance while neglecting tail instances. To remedy this, the authors propose DyCA, a framework that incorporates dynamic, feature-agnostic instance clustering as a co-evolutionary component. By identifying groups of instances exhibiting similar algorithmic response patterns, DyCA enables structure-aware, fine-grained algorithm optimization. Integrating LLM-guided evolutionary search with algorithm composition, the approach achieves a 15.2% improvement in average tail performance and a 7.1% gain in overall performance across four heterogeneous algorithm design tasks, all while maintaining state-of-the-art head performance.

Abstract

Large Language Model-assisted Evolutionary Search (LES) has emerged as a powerful paradigm for automated algorithm design. However, existing LES methods primarily optimize for average performance, inherently directing search effort toward instances that contribute most to this metric while leaving others poorly served, resulting in weak tail robustness and limited real-world reliability. To address this limitation, we propose Dynamic Instance Clustering and Specialized Algorithm Design (DyCA), an LES framework with a feature-free, structure-aware mechanism for constructing reliable algorithm portfolios under heterogeneous instance distributions. DyCA treats instance clustering as a co-evolving component within the search process, reusing accumulated evaluation data as feature-free signals to progressively partition instances with similar algorithmic response patterns. The uncovered clusters decompose the mixed objective into a set of structure-aware sub-objectives, thereby enabling finer-grained and more adaptive guidance for specialized algorithm design. Experimental results across four algorithm design tasks with heterogeneous instances demonstrate that DyCA outperforms state-of-the-art LES baselines, improving tail robustness by an average of 15.2\% and overall performance by 7.1\% while maintaining competitive head performance.

Problem

• tail robustness
• heterogeneous instances
• algorithm design
• evolutionary search
• LLM-assisted optimization

Innovation

• Dynamic Instance Clustering
• Specialized Algorithm Design
• Feature-free Clustering
• Tail Robustness
• LLM-assisted Evolutionary Search