Abstract:Heuristics are widely used for dealing with complex search and optimization problems. However, manual design of heuristics can be often very labour extensive and requires rich working experience and knowledge. This paper proposes Evolution of Heuristic (EoH), a novel evolutionary paradigm that leverages both Large Language Models (LLMs) and Evolutionary Computation (EC) methods for Automatic Heuristic Design (AHD). EoH represents the ideas of heuristics in natural language, termed thoughts. They are then translated into executable codes by LLMs. The evolution of both thoughts and codes in an evolutionary search framework makes it very effective and efficient for generating high-performance heuristics. Experiments on three widely studied combinatorial optimization benchmark problems demonstrate that EoH outperforms commonly used handcrafted heuristics and other recent AHD methods including FunSearch. Particularly, the heuristic produced by EoH with a low computational budget (in terms of the number of queries to LLMs) significantly outperforms widely-used human hand-crafted baseline algorithms for the online bin packing problem.

What problem does this paper attempt to address?

The paper aims to address the problem of automatically designing efficient heuristic algorithms, particularly in complex search and optimization problems where manually designing heuristic algorithms is both time-consuming and requires extensive expertise. The paper proposes a new framework called "Evolution of Heuristics" (EoH), which combines large language models (LLMs) with evolutionary computation methods to automatically design heuristic algorithms. ### Research Objectives - **Reduce Human Intervention**: Reduce the need for expert manual design of heuristic algorithms through an automated process. - **Increase Efficiency**: Improve the efficiency of generating high-quality heuristic algorithms compared to existing methods. - **Generality**: The proposed method can be applied to various combinatorial optimization problems. ### Main Contributions 1. **EoH Framework**: Proposes the EoH framework, which uses LLMs to evolve both the natural language descriptions (ideas) and code implementations of heuristics, achieving automatic heuristic design with minimal human design and no need for domain-specific model training. 2. **Prompt Strategies**: Develops several simple yet effective prompt strategies to guide LLMs in generating more diverse and effective heuristic algorithms. These strategies are generally applicable to other LLM-assisted search methods. 3. **Experimental Validation**: Conducts comprehensive evaluations of EoH on three widely studied combinatorial optimization benchmark problems, demonstrating its performance superiority over many existing automatic heuristic design methods. Specifically, for the online bin packing problem, the heuristics identified by EoH outperform those generated by methods like FunSearch, with significantly fewer LLM queries required. ### Method Overview - **Heuristic Representation**: Each heuristic algorithm includes a natural language description, code block, and fitness value. - **Evolutionary Framework**: EoH maintains a population of heuristic algorithms, using several prompt strategies (including exploration and modification strategies) to generate new heuristics and guide the search direction through selection operations. - **Integrating LLMs**: Utilizes LLMs to generate natural language descriptions and corresponding code implementations of heuristics, continuously improving these heuristics through the evolutionary process. ### Experimental Results - For the online bin packing problem, the heuristics generated by EoH significantly outperform manually designed heuristics (such as Best Fit) and automatic design methods like FunSearch, especially with lower computational resource consumption. - For the Traveling Salesman Problem (TSP), EoH also shows excellent performance, particularly in terms of the gap to the known optimal solution. - In the flow shop scheduling problem, EoH performs outstandingly, generating heuristics that surpass traditional manually designed heuristics. In summary, the paper effectively addresses the problem of automatically designing heuristic algorithms by proposing the EoH framework and validates its superiority across multiple combinatorial optimization problems.

Evolution of Heuristics: Towards Efficient Automatic Algorithm Design Using Large Language Model

Understanding the Importance of Evolutionary Search in Automated Heuristic Design with Large Language Models

Multi-objective Evolution of Heuristic Using Large Language Model

Deep Insights into Automated Optimization with Large Language Models and Evolutionary Algorithms

ReEvo: Large Language Models as Hyper-Heuristics with Reflective Evolution

Algorithm Evolution Using Large Language Model.

HSEvo: Elevating Automatic Heuristic Design with Diversity-Driven Harmony Search and Genetic Algorithm Using LLMs

UBER: Uncertainty-Based Evolution with Large Language Models for Automatic Heuristic Design

HeurAgenix: A Multi-Agent LLM-Based Paradigm for Adaptive Heuristic Evolution and Selection in Combinatorial Optimization

Large Language Models as Evolutionary Optimizers

Learning-aided Evolution for Optimization

Automatic programming via large language models with population self-evolution for dynamic job shop scheduling problem

Artificial Evolutionary Intelligence (AEI): Evolutionary Computation Evolves with Large Language Models

Large Language Model Aided Multi-objective Evolutionary Algorithm: a Low-cost Adaptive Approach

LLaMEA: A Large Language Model Evolutionary Algorithm for Automatically Generating Metaheuristics

Large Language Model-Aided Evolutionary Search for Constrained Multiobjective Optimization

Automated Design of Collaboration-Based Hybrid Metaheuristics

Automated Design of Search Algorithms Based on Reinforcement Learning.

Human Evolutionary Optimization Algorithm

A novel evolutionary status guided hyper-heuristic algorithm for continuous optimization

Evolving Real-time Heuristic Search Algorithms