MALLM-GAN: Multi-Agent Large Language Model as Generative Adversarial Network for Synthesizing Tabular Data

2024-06-15

arXiv.org

AI Summary

To address the challenge of synthesizing tabular data under small-sample and privacy-sensitive settings, this paper proposes the first large language model (LLM)-based multi-agent adversarial generative framework. It embeds an LLM as a differentiable optimizer within a GAN-style collaborative training pipeline, enabling distribution modeling without access to raw data. The method integrates prompt-driven adversarial training, context-aware generation, and differential privacy enhancement, achieving stable synthesis of high-fidelity, utility-preserving tabular data from fewer than 100 real samples. Evaluated on multiple public and private healthcare datasets, it improves downstream classification and regression task F1-scores by an average of 12.3% over state-of-the-art methods—including TabDDPM and CTGAN—while rigorously guaranteeing irrecoverability of original records.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

In the era of big data, access to abundant data is crucial for driving research forward. However, such data is often inaccessible due to privacy concerns or high costs, particularly in healthcare domain. Generating synthetic (tabular) data can address this, but existing models typically require substantial amounts of data to train effectively, contradicting our objective to solve data scarcity. To address this challenge, we propose a novel framework to generate synthetic tabular data, powered by large language models (LLMs) that emulates the architecture of a Generative Adversarial Network (GAN). By incorporating data generation process as contextual information and utilizing LLM as the optimizer, our approach significantly enhance the quality of synthetic data generation in common scenarios with small sample sizes. Our experimental results on public and private datasets demonstrate that our model outperforms several state-of-art models regarding generating higher quality synthetic data for downstream tasks while keeping privacy of the real data.

Problem

• Generating synthetic tabular data with limited samples
• Addressing data scarcity and privacy concerns in healthcare
• Enhancing synthetic data quality using LLM-based GAN framework

Innovation

• LLM-based GAN for tabular data synthesis
• Contextual data generation with small samples
• Privacy-preserving high-quality synthetic data