SYNTHONY: A Stress-Aware, Intent-Conditioned Agent for Deep Tabular Generative Models Selection

TL;DR

SYNTHONY is a stress-aware, intent-conditioned framework for selecting the most suitable deep tabular data synthesizer based on dataset difficulty and user preferences, improving selection accuracy.

Contribution

The paper introduces stress profiling and a selection framework that effectively matches dataset stress profiles to synthesizer capabilities, advancing data synthesis selection methods.

Findings

Stress-based meta-features predict synthesizer performance accurately.

The $k$NN selector outperforms zero-shot LLM and random baselines.

Identifies the need for learned capability representations for further improvement.

Abstract

Deep generative models for tabular data (GANs, diffusion models, and LLM-based generators) exhibit highly non-uniform behavior across datasets; the best-performing synthesizer family depends strongly on distributional stressors such as long-tailed marginals, high-cardinality categorical, Zipfian imbalance, and small-sample regimes. This brittleness makes practical deployment challenging, especially when users must balance competing objectives of fidelity, privacy, and utility. We study {intent-conditioned tabular synthesis selection}: given a dataset and a user intent expressed as a preference over evaluation metrics, the goal is to select a synthesizer that minimizes regret relative to an intent-specific oracle. We propose {stress profiling}, a synthesis-specific meta-feature representation that quantifies dataset difficulty along four interpretable stress dimensions, and integrate it into {SYNTHONY}, a selection framework that matches stress profiles against a calibrated capability registry of synthesizer families. Across a benchmark of 7 datasets, 10 synthesizers, and 3 intents, we demonstrate that stress-based meta-features are highly predictive of synthesizer performance: a $k$NN selector using these features achieves strong Top-1 selection accuracy, substantially outperforming zero-shot LLM selectors and random baselines. We analyze the gap between meta-feature-based and capability-based selection, identifying the hand-crafted capability registry as the primary bottleneck and motivating learned capability representations as a direction for future work.