Research map / Probabilistic and scientific modeling / Variational and generative sampling
TabGeoFlow: A Geometric Flow Matching Model for Tabular Data Synthesis
Jong In Choi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add structural flow regularization (SFR) loss (ablation) | TabGeoFlow · Adult | TSTR AUC: .898 -> .913 | Improved |
| add structural flow regularization (SFR) loss (ablation) | TabGeoFlow · Default | TSTR AUC: .745 -> .761 | Improved |
| add SFR loss (ablation) | TabGeoFlow · Adult / Default | DCR % (distance from 50): Adult 51.11 -> 50.37; Default 48.46 -> 50.19 | Mixed |
| use dynamic threshold instead of fixed (ablation) | TabGeoFlow · Adult | TSTR AUC: .903 -> .913 | Improved |
| use dynamic threshold instead of fixed (ablation) | TabGeoFlow · Default | TSTR AUC: .760 -> .761 | No clear effect |
| vary tangent dimension k (10, 20, 40 vs 30) (ablation) | TabGeoFlow · Adult | TSTR AUC: .903-.909 vs .913 | No clear effect |
| use TabGeoFlow instead of TabDiff for privacy | tabular synthesizer · DCR average over 7 datasets | DCR %: 52.06 -> 50.45 | Improved |
| use TabGeoFlow instead of TabDiff for fidelity | tabular synthesizer · Shape error average | shape error rate %: 1.24 -> 1.34 | No clear effect |
Closest papers by meaning
- Breaking the Quality–Privacy Tradeoff in Tabular Data Generation via In-Context Learning (NeurIPS 2026 · similarity 0.88)
- Exponential Family Variational Flow Matching for Tabular Data Generation (ICML 2025 · similarity 0.88)
- TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation (ICLR 2025 · similarity 0.86)
- CTSyn: A Foundation Model for Cross Tabular Data Generation (ICLR 2025 · similarity 0.86)
- CuTS: Customizable Tabular Synthetic Data Generation (ICML 2024 · similarity 0.86)
- TabSDS: a Lightweight, Fully Non-Parametric, and Model Free Approach for Generating Synthetic Tabular Data (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
