Inference-time refinement of pre-trained tabular diffusion models via Bidirectional Chamfer Refinement achieves median 8.6% better downstream performance than real data across 15 benchmarks while preserving fidelity and privacy.
hub
Mixed-type tabular data synthesis with score-based diffusion in latent space
11 Pith papers cite this work. Polarity classification is still indexing.
11
Pith papers citing it
hub tools
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 11roles
background 1polarities
background 1representative citing papers
FERMI improves membership inference on tabular diffusion models by mapping relational auxiliary information into attack features, raising TPR at 0.1 FPR by up to 53% white-box and 22% black-box over single-table baselines.
GRDM jointly generates relational database tables via graph-conditional diffusion without table ordering, outperforming autoregressive baselines on multi-hop correlations and single-table fidelity across six real RDBs.
LLM-TabLogic extracts inter-column logical constraints using LLMs and conditions a score-based latent diffusion model on them to generate synthetic tabular data that preserves those relationships.
DiffICL breaks the quality-privacy tradeoff in small-data tabular synthesis by using in-context learning on pretrained structural priors to generate data that is both higher quality and less memorizing of training samples.
RDDG is an in-context learning system with dynamic guidance and automatic quality feedback that synthesizes high-fidelity relational data to improve imbalanced classification.
Proposes three metrics for inter-column logical relationships in synthetic tabular data and reports that current generators often fail to preserve them on an industrial dataset.
TabGRAA applies group-relative advantage alignment in an iterative reward-guided post-training loop to improve tabular language model generators on fidelity, utility, and privacy trade-offs across five benchmarks.
A temporal extension of TabDDPM generates coherent synthetic time-series sequences on the WISDM dataset that match real distributions and support downstream classification with macro F1 of 0.64.
A new end-to-end benchmarking framework unifies synthetic EHR generators for longitudinal ICD codes with standardized training and architecture-agnostic evaluation including bootstrapped confidence intervals.
The MIDST challenge evaluated privacy resilience of diffusion-generated synthetic tabular data via membership inference attacks and produced new black-box and white-box attack methods.
citing papers explorer
-
Inference-Time Refinement Closes the Synthetic-Real Gap in Tabular Diffusion
Inference-time refinement of pre-trained tabular diffusion models via Bidirectional Chamfer Refinement achieves median 8.6% better downstream performance than real data across 15 benchmarks while preserving fidelity and privacy.
-
FERMI: Exploiting Relations for Membership Inference Against Tabular Diffusion Models
FERMI improves membership inference on tabular diffusion models by mapping relational auxiliary information into attack features, raising TPR at 0.1 FPR by up to 53% white-box and 22% black-box over single-table baselines.
-
Joint Relational Database Generation via Graph-Conditional Diffusion Models
GRDM jointly generates relational database tables via graph-conditional diffusion without table ordering, outperforming autoregressive baselines on multi-hop correlations and single-table fidelity across six real RDBs.
-
LLM-TabLogic: Preserving Inter-Column Logical Relationships in Synthetic Tabular Data via Prompt-Guided Latent Diffusion
LLM-TabLogic extracts inter-column logical constraints using LLMs and conditions a score-based latent diffusion model on them to generate synthetic tabular data that preserves those relationships.
-
Breaking the Quality-Privacy Tradeoff in Tabular Data Generation via In-Context Learning
DiffICL breaks the quality-privacy tradeoff in small-data tabular synthesis by using in-context learning on pretrained structural priors to generate data that is both higher quality and less memorizing of training samples.
-
Self-Reinforcing Controllable Synthesis of Rare Relational Data via Bayesian Calibration
RDDG is an in-context learning system with dynamic guidance and automatic quality feedback that synthesizes high-fidelity relational data to improve imbalanced classification.
-
Evaluating Inter-Column Logical Relationships in Synthetic Tabular Data Generation
Proposes three metrics for inter-column logical relationships in synthetic tabular data and reports that current generators often fail to preserve them on an industrial dataset.
-
Self-Improving Tabular Language Models via Iterative Reward-Guided Post-Training
TabGRAA applies group-relative advantage alignment in an iterative reward-guided post-training loop to improve tabular language model generators on fidelity, utility, and privacy trade-offs across five benchmarks.
-
Extending Tabular Denoising Diffusion Probabilistic Models for Time-Series Data Generation
A temporal extension of TabDDPM generates coherent synthetic time-series sequences on the WISDM dataset that match real distributions and support downstream classification with macro F1 of 0.64.
-
Accelerating Reproducible Research in Synthetic EHR Generation
A new end-to-end benchmarking framework unifies synthetic EHR generators for longitudinal ICD codes with standardized training and architecture-agnostic evaluation including bootstrapped confidence intervals.
-
MIDST Challenge at SaTML 2025: Membership Inference over Diffusion-models-based Synthetic Tabular data
The MIDST challenge evaluated privacy resilience of diffusion-generated synthetic tabular data via membership inference attacks and produced new black-box and white-box attack methods.