HiFo-Prompt: Prompting with Hindsight and Foresight for LLM-based Automatic Heuristic Design

2025-08-18

AI Summary

Large language models (LLMs) face two key bottlenecks in automated heuristic design (AHD) for evolutionary computation: over-reliance on static operators and inability to accumulate domain-specific knowledge across optimization runs. Method: We propose a feedforward–backtracking dual-phase collaborative prompting framework. It leverages population dynamics analysis to drive adaptive prompt engineering and integrates an experience replay mechanism that distills historically successful strategies into transferable, general-purpose heuristic principles—enabling continuous self-improvement of the LLM during search. Contribution/Results: The resulting knowledge accumulation system effectively balances exploration and exploitation, overcoming limitations of fixed operators and catastrophic forgetting. Empirical evaluation demonstrates that our approach generates superior heuristics with significantly fewer LLM queries—achieving faster convergence and up to several-fold improvement in query efficiency compared to baseline methods.

Technology Category

• Search and Optimization: Heuristic Search
• Machine Learning: Evolutionary Learning
• Natural Language Processing: Prompt Engineering / Prompting

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

LLM-based Automatic Heuristic Design (AHD) within Evolutionary Computation (EC) frameworks has shown promising results. However, its effectiveness is hindered by the use of static operators and the lack of knowledge accumulation mechanisms. We introduce HiFo-Prompt, a framework that guides LLMs with two synergistic prompting strategies: Foresight and Hindsight. Foresight-based prompts adaptively steer the search based on population dynamics, managing the exploration-exploitation trade-off. In addition, hindsight-based prompts mimic human expertise by distilling successful heuristics from past generations into fundamental, reusable design principles. This dual mechanism transforms transient discoveries into a persistent knowledge base, enabling the LLM to learn from its own experience. Empirical results demonstrate that HiFo-Prompt significantly outperforms state-of-the-art LLM-based AHD methods, generating higher-quality heuristics while achieving substantially faster convergence and superior query efficiency.

Problem

• Addresses static operators in LLM-based heuristic design
• Overcomes lack of knowledge accumulation in evolutionary computation
• Enhances quality and efficiency of automatic heuristic generation

Innovation

• Adaptive prompting based on population dynamics
• Distilling past heuristics into reusable principles
• Dual mechanism for persistent knowledge base