Leveraging LLM Agents for Automated Optimization Modeling for SASP Problems: A Graph-RAG based Approach

2025-01-30

AI Summary

In automated optimization modeling (AOM) for sensor array signal processing (SASP), inaccurate modeling arises from insufficient domain knowledge integration. Method: This paper proposes a human-modeling-principle-driven multi-agent collaborative framework augmented with graph-structured retrieval-augmented generation (Graph-RAG). It introduces the first multi-agent architecture specifically designed for the SASP modeling pipeline, leveraging graph neural networks to enhance knowledge retrieval and enable precise alignment between user queries and fine-grained SASP domain knowledge. Contribution/Results: Evaluated on ten canonical SASP tasks, our approach substantially outperforms existing AOM methods, achieving state-of-the-art accuracy in both modeling correctness and executable code generation. It marks the first successful integration of deep domain knowledge embedding with large language model–based reasoning and planning, enabling organic synergy between expert modeling principles and data-driven inference.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Automated optimization modeling (AOM) has evoked considerable interest with the rapid evolution of large language models (LLMs). Existing approaches predominantly rely on prompt engineering, utilizing meticulously designed expert response chains or structured guidance. However, prompt-based techniques have failed to perform well in the sensor array signal processing (SASP) area due the lack of specific domain knowledge. To address this issue, we propose an automated modeling approach based on retrieval-augmented generation (RAG) technique, which consists of two principal components: a multi-agent (MA) structure and a graph-based RAG (Graph-RAG) process. The MA structure is tailored for the architectural AOM process, with each agent being designed based on principles of human modeling procedure. The Graph-RAG process serves to match user query with specific SASP modeling knowledge, thereby enhancing the modeling result. Results on ten classical signal processing problems demonstrate that the proposed approach (termed as MAG-RAG) outperforms several AOM benchmarks.

Problem

• Automated Optimization Modeling
• Sensor Array Signal Processing
• Domain Expertise Lack

Innovation

• MAG-RAG
• Graph-RAG
• Sensor Array Signal Processing