ReTabSyn: Realistic Tabular Data Synthesis via Reinforcement Learning

TL;DR

ReTabSyn introduces a reinforcement learning-based method for synthesizing realistic tabular data by focusing on conditional distributions, improving data utility especially in low-data, imbalanced, and distribution-shifted scenarios.

Contribution

The paper proposes ReTabSyn, a novel reinforcement learning pipeline that emphasizes learning conditional distributions for more effective tabular data synthesis.

Findings

ReTabSyn outperforms state-of-the-art baselines on benchmarks with small sample sizes.

The method effectively handles class imbalance and distribution shift.

It can incorporate expert constraints into synthetic data generation.

Abstract

Deep generative models can help with data scarcity and privacy by producing synthetic training data, but they struggle in low-data, imbalanced tabular settings to fully learn the complex data distribution. We argue that striving for the full joint distribution could be overkill; for greater data efficiency, models should prioritize learning the conditional distribution $P(y\mid \bm{X})$, as suggested by recent theoretical analysis. Therefore, we overcome this limitation with \textbf{ReTabSyn}, a \textbf{Re}inforced \textbf{Tab}ular \textbf{Syn}thesis pipeline that provides direct feedback on feature correlation preservation during synthesizer training. This objective encourages the generator to prioritize the most useful predictive signals when training data is limited, thereby strengthening downstream model utility. We empirically fine-tune a language model-based generator using this approach, and across benchmarks with small sample sizes, class imbalance, and distribution shift, ReTabSyn consistently outperforms state-of-the-art baselines. Moreover, our approach can be readily extended to control various aspects of synthetic tabular data, such as applying expert-specified constraints on generated observations.