Abstract:Optimization algorithms and large language models (LLMs) enhance decision-making in dynamic environments by integrating artificial intelligence with traditional techniques. LLMs, with extensive domain knowledge, facilitate intelligent modeling and strategic decision-making in optimization, while optimization algorithms refine LLM architectures and output quality. This synergy offers novel approaches for advancing general AI, addressing both the computational challenges of complex problems and the application of LLMs in practical scenarios. This review outlines the progress and potential of combining LLMs with optimization algorithms, providing insights for future research directions.

What problem does this paper attempt to address?

This paper focuses on the combination of Large Language Models (LLMs) and optimization algorithms to solve decision optimization problems in dynamic environments. The authors point out that optimization algorithms use biological evolution mechanisms to deal with complex problems, while LLMs facilitate intelligent modeling and strategic decision-making with their rich domain knowledge. By combining the two, new paths can be opened for the development of general artificial intelligence, while addressing the computational challenges of complex problems and the implementation of LLMs in practical applications. The paper introduces breakthroughs of LLMs such as GPT in natural language understanding and generation, and discusses their wide applications in multiple fields including logistics, finance, healthcare, and artificial intelligence. However, training and applying LLMs require significant resources, and optimization algorithms can improve the efficiency and output quality of model structures. The article reviews the history of optimization algorithms, from iterative methods to advanced algorithms that adapt to different problems and constraints. Reinforcement learning is also mentioned as a learning strategy that optimizes long-term rewards through interaction with the environment. In addition, the paper explores classical reinforcement learning and deep reinforcement learning methods such as Q-learning, REINFORCE, TRPO, and PPO. The main goal of the paper is to systematically analyze the development and application of LLMs and optimization algorithms, summarize related research, and explore future research directions. It emphasizes the importance of designing optimization algorithms to improve LLMs efficiency and reduce resource consumption while maintaining or enhancing model performance. Finally, the paper discusses the practical applications of integrating optimization algorithms with LLMs and prospects future research trends.

When Large Language Model Meets Optimization

When large language model meets optimization

Towards Optimizing with Large Language Models

Exploring the True Potential: Evaluating the Black-box Optimization Capability of Large Language Models

Solving General Natural-Language-Description Optimization Problems with Large Language Models

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

Autonomous Multi-Objective Optimization Using Large Language Model

Position: Leverage Foundational Models for Black-Box Optimization

A Systematic Survey on Large Language Models for Algorithm Design

Metaheuristics and Large Language Models Join Forces: Towards an Integrated Optimization Approach

Large Language Model-Based Evolutionary Optimizer: Reasoning with elitism

Towards Next Era of Multi-objective Optimization: Large Language Models as Architects of Evolutionary Operators

Using Large Language Models for Hyperparameter Optimization

Optimal Decision Making Through Scenario Simulations Using Large Language Models

Large Language Models for Constructing and Optimizing Machine Learning Workflows: A Survey

Large Language Models as Evolutionary Optimizers

Exploring the landscape of large language models: Foundations, techniques, and challenges

Adaptive Optimization for Enhanced Efficiency in Large-Scale Language Model Training

Large Language Models for Combinatorial Optimization of Design Structure Matrix

Exploring the Improvement of Evolutionary Computation via Large Language Models