Research map / Image generation / Fast diffusion sampling
Active Tabular Augmentation via Policy-Guided Diffusion Inpainting
Zheyu Zhang, Shuo Yang, Bardh Prenkaj, Gjergji Kasneci
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 point-wise feasibility gating (ablation) | TAP · Safe injection ablation (n_real 20/50/100) | win-rate: 57.3% -> 100.0% | Improved |
| add point-wise feasibility gating (ablation) | TAP · Safe injection ablation (n_real 20/50/100) | tail risk: 61.9% -> 46.7% | Improved |
| add point-wise gating (ablation) | TAP · Safe injection ablation (n_real=50) | utility gain: 0.037 -> 0.095 | Improved |
| add point-wise gating (ablation) | TAP · Safe injection ablation (n_real=20) | utility gain: 0.108 -> 0.140 | Improved |
| add windowed commitment (ablation) | TAP · Safe injection ablation (n_real 20/50/100) | win-rate: 85.3% -> 100.0% | Improved |
| add windowed commitment (ablation) | TAP · Safe injection ablation (n_real 20/50/100) | tail risk: 47.2% -> 46.7% | No clear effect |
| add windowed commitment (ablation) | TAP · Safe injection ablation (n_real=100) | utility gain: 0.085 -> 0.097 | Improved |
| use TAP instead of TabDiff augmentation | TabDiff backbone · Steel (n_real=50, classification) | accuracy (%): 87.81 -> 94.27 | Improved |
| use TAP instead of TabDiff augmentation | TabDiff backbone · MiceProtein (n_real=20, classification) | accuracy (%): 37.85 -> 44.60 | Improved |
| use generative augmentation TabDiff instead of real data only | Downstream predictors · Electricity (n_real=20, classification) | accuracy (%): 66.09 -> 60.17 | Hurt |
| use TabDDPM augmentation instead of real data only | Downstream regressors · Ailerons (n_real=100, regression) | RMSE: 0.596 -> 108.419 | Hurt |
| use TAP instead of TabDiff augmentation | TabDiff backbone · Insurance (n_real=20, regression) | RMSE: 1.067 -> 0.885 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TabDiff — TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation (ICLR 2025) · better on 8 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation (ICLR 2025 · similarity 0.87)
- Tabular Foundation Model for Generative Modelling (NeurIPS 2026 · similarity 0.86)
- Understanding and Mitigating Memorization in Diffusion Models for Tabular Data (ICML 2025 · similarity 0.86)
- TabNAT: A Continuous-Discrete Joint Generative Framework for Tabular Data (ICML 2025 · similarity 0.86)
- Not Just Pretty Pictures: Toward Interventional Data Augmentation Using Text-to-Image Generators (ICML 2024 · similarity 0.85)
- Data Augmentation with Diffusion for Open-Set Semi-Supervised Learning (NeurIPS 2024 · similarity 0.85)
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.
