# Inference-Time Refinement Closes the Synthetic-Real Gap in Tabular Diffusion

arXiv: https://arxiv.org/abs/2605.06261 · HTML: https://mlsanity.com/paper/2605.06261/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 24 references that are in the map.
- First posted on arXiv 7 May 2026.

## Builds on (most cited first, 17 of 17)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Classifier-Free Diffusion Guidance](https://mlsanity.com/paper/2207.12598/) (2022)
- [XGBoost: A Scalable Tree Boosting System](https://mlsanity.com/paper/1603.02754/) (2016)
- [Improved Precision and Recall Metric for Assessing Generative Models](https://mlsanity.com/paper/1904.06991/) (2019)
- [Modeling Tabular data using Conditional GAN](https://mlsanity.com/paper/1907.00503/) (2019)
- [TabDDPM: Modelling Tabular Data with Diffusion Models](https://mlsanity.com/paper/2209.15421/) (2022)
- [Mixed-Type Tabular Data Synthesis with Score-based Diffusion in Latent Space](https://mlsanity.com/paper/2310.09656/) (2023)
- [STaSy: Score-based Tabular data Synthesis](https://mlsanity.com/paper/2210.04018/) (2022)
- [CoDi: Co-evolving Contrastive Diffusion Models for Mixed-type Tabular Synthesis](https://mlsanity.com/paper/2304.12654/) (2023)
- [CTAB-GAN+: Enhancing Tabular Data Synthesis](https://mlsanity.com/paper/2204.00401/) (2022)
- [TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation](https://mlsanity.com/paper/2410.20626/) (2024)
- [Information-Based Optimal Subdata Selection for Big Data Linear Regression](https://mlsanity.com/paper/1710.10382/) (2017)
- [Navigating Tabular Data Synthesis Research: Understanding User Needs and Tool Capabilities](https://mlsanity.com/paper/2405.20959/) (2024)
- [Federated Knowledge Recycling: Privacy-Preserving Synthetic Data Sharing](https://mlsanity.com/paper/2407.20830/) (2024)
- [Increasing the Utility of Synthetic Images through Chamfer Guidance](https://mlsanity.com/paper/2508.10631/) (2025)
- [Synthetic Image Learning: Preserving Performance and Preventing Membership Inference Attacks](https://mlsanity.com/paper/2407.15526/) (2024)

## Built on by (most cited first, 1 of 1)

- [A Filtered Mixture-of-Generators for Fully Synthetic Survival Training](https://mlsanity.com/paper/2607.00127/) (2026)

## Abstract (arXiv)

Diffusion-based generators set the current state of the art for synthetic tabular data. These methods approach but rarely exceed real-data utility, and closing this synthetic-real gap has so far been pursued exclusively at training time, via architectural advances, scaling, and retraining of monolithic generators. The inference-time alternative, i.e., refining the outputs of a pre-trained backbone with parameters left untouched, has remained largely unexplored for tabular synthesis. We introduce TARDIS (Tabular generation through Refinement, Distillation, and Inference-time Sampling), an inference-time refinement framework that operates on a frozen pre-trained backbone, configured per dataset by a Tree-structured Parzen Estimator search over score-level guidance during reverse diffusion, with each trial's objective set by an inner grid search over post-hoc sample selectors and an optional soft-label distillation step. The search space encodes a single mathematical pattern we name Bidirectional Chamfer Refinement (BCR): the symmetric Chamfer functional between synthetic and real samples is minimized both continuously, via a score-level gradient, and discretely, via batch-ranking post-generation. The per-dataset search recovers BCR-aligned configurations on most datasets, evidence for BCR as the dominant refinement pattern. Across 15 binary, multiclass, and regression benchmarks TARDIS achieves a median +8.6% downstream-task improvement over models trained on real data (95% CI [+3.3, +16.4], Wilcoxon p=0.016, 11/15 strict wins) and improves over the TabDiff backbone on all 15 datasets (mean +12.9%, p<10^-4), matching the backbone on manifold fidelity, diversity, and sample-level privacy. Inference-time refinement of a pre-trained tabular diffusion backbone reaches and exceeds real-data utility in 1 to 80 minutes on a single consumer-grade GPU.