Dependency-aware synthetic tabular data generation

2025-07-25

AI Summary

Existing synthetic tabular data generation models struggle to preserve functional dependencies (FDs) and logical dependencies (LDs) among attributes, leading to structural distortions and limiting their applicability in privacy-sensitive domains such as healthcare. To address this, we propose the Hierarchical Feature Generation Framework (HFGF), a model-agnostic approach that explicitly encodes and incorporates FD/LD constraints without modifying underlying generative models (e.g., CTGAN, TVAE, GReaT). HFGF reconstructs data in stages—first generating independent features, then conditionally generating dependent ones—thereby enabling structure-aware synthesis. Experiments across four benchmark datasets demonstrate that HFGF significantly enhances the FD/LD preservation capability of six state-of-the-art models. The resulting synthetic data exhibit improved structural consistency and superior downstream performance in classification and regression tasks. This work establishes a new paradigm for high-fidelity, interpretable synthetic data generation.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Feature Construction/Reformulation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Synthetic tabular data is increasingly used in privacy-sensitive domains such as health care, but existing generative models often fail to preserve inter-attribute relationships. In particular, functional dependencies (FDs) and logical dependencies (LDs), which capture deterministic and rule-based associations between features, are rarely or often poorly retained in synthetic datasets. To address this research gap, we propose the Hierarchical Feature Generation Framework (HFGF) for synthetic tabular data generation. We created benchmark datasets with known dependencies to evaluate our proposed HFGF. The framework first generates independent features using any standard generative model, and then reconstructs dependent features based on predefined FD and LD rules. Our experiments on four benchmark datasets with varying sizes, feature imbalance, and dependency complexity demonstrate that HFGF improves the preservation of FDs and LDs across six generative models, including CTGAN, TVAE, and GReaT. Our findings demonstrate that HFGF can significantly enhance the structural fidelity and downstream utility of synthetic tabular data.

Problem

• Preserving inter-attribute relationships in synthetic tabular data
• Addressing poor retention of functional and logical dependencies
• Enhancing structural fidelity of synthetic data for downstream utility

Innovation

• Hierarchical Feature Generation Framework for tabular data
• Reconstructs dependent features using FD and LD rules
• Improves preservation of functional and logical dependencies