Exploring the Heterogeneity of Tabular Data: A Diversity-aware Data Generator via LLMs

TL;DR

This paper introduces DATE, a diversity-aware data generator using LLMs that partitions heterogeneous data and balances diversity with quality, significantly improving tabular data generation for machine learning tasks.

Contribution

The paper proposes a novel framework DATE that effectively partitions data, explores diversity with LLMs, and employs a Multi-Arm Bandit algorithm to balance diversity and quality, outperforming existing methods.

Findings

Achieves 23.75% reduction in error rate on benchmarks.

Improves accuracy of DPO and reasoning of LLMs.

Outperforms state-of-the-art GAN and LLM-based data generators.

Abstract

Tabular data generation has become increasingly essential for enabling robust machine learning applications, which require large-scale, high-quality data. Existing solutions leverage generative models to learn original data distributions. However, real-world data are naturally heterogeneous with diverse distributions, making it challenging to obtain a universally good model for diverse data generation. To address this limitation, we introduce Diversity-Aware Tabular data gEnerator (DATE), a framework that (i) prepares high-quality and distributionally distinct examples for in-context learning by effectively partitioning the original heterogeneous data into multiple diverse subsets; (ii) harnesses Large Language Models (LLMs) to explore the diversity of the partitioned distribution with decision tree reasoning as feedback, generating high-quality labeled data for each subset. However, the massive generated data inherently involves a trade-off between diversity and quality. To integrate this issue, existing solutions greedily select the validation-best data. However, we prove that the selection in heterogeneous settings does not possess the greedy-choice property, and design a Multi-Arm Bandit-based sampling algorithm that balances the diversity and quality of generated data. Extensive experiments on tabular classification and regression benchmarks demonstrate that DATE consistently outperforms state-of-the-art GAN-based and LLM-based methods. On average, DATE achieves a 23.75% reduction in error rate with just 100 generated data. Empirically, we demonstrate that data generated by DATE can improve the accuracy of Direct Preference Optimization (DPO) and enhance the reasoning capability of LLMs on the target data. Code is available at https://github.com/windblow32/DATE.