Research map / Architectures and efficiency / RNNs and normalization
Geometry-Aware Tabular Diffusion
David Turtora Zagardo
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 supervised geometric losses (angle, length, consistency) to MLP denoiser (ablation) | Residual MLP diffusion · 10 tabular datasets | Shape error: 1.279 -> 0.862 (InputsOnly -> Full) | Improved |
| add geometric inputs and heads without supervision (ablation) | Residual MLP diffusion · 10 tabular datasets | Shape error: 1.229 -> 1.279 (d = -0.08) | No clear effect |
| add geometric supervision to GNN denoiser | GNN+LE diffusion · 10 tabular datasets | Shape error: 1.23 -> 1.17 | Mixed |
| add geometric supervision to Transformer denoiser | Column-wise Transformer diffusion · 10 tabular datasets | Shape error: 1.23 -> 0.959 | Improved |
| add geometric supervision to GNN denoiser | GNN+LE diffusion · 10 tabular datasets | Trend error: 5.72 -> 4.19 | Improved |
| replace TabDiff transformer with MLP+Geom | TabDiff · 10 tabular datasets | Shape error: 1.19 -> 0.862 | Improved |
| replace TabDiff transformer with MLP+Geom | TabDiff · 10 tabular datasets | MLE-1 gap to real: 11.4% -> 4.8% | Improved |
| match TabDiff training budget at 8,000 epochs (ablation) | GATD-MLP · 10 tabular datasets | Shape error: 0.862 -> 0.854 | No clear effect |
| reduce sampling steps from 1000 to 50 (ablation) | GATD-MLP · 10 tabular datasets | Shape error: 0.862 -> 1.546 | Hurt |
| reduce sampling steps from 1000 to 100 (ablation) | GATD-MLP · 10 tabular datasets | Shape error: 0.862 -> 1.125 | Hurt |
| use geometric supervision on California with MLP | GATD-MLP · California | Shape error: 22.8 percentage points worse than TabDiff | Mixed |
| raise geometric loss weights to dominate diffusion loss (ablation) | GATD · 10 tabular datasets | fidelity | Improved |
Closest papers by meaning
- TabDiff: a Mixed-type Diffusion Model for Tabular Data Generation (ICLR 2025 · similarity 0.90)
- Continuous Diffusion for Mixed-Type Tabular Data (ICLR 2025 · similarity 0.88)
- ClavaDDPM: Multi-relational Data Synthesis with Cluster-guided Diffusion Models (NeurIPS 2024 · similarity 0.87)
- D2R2: Diffusion-based Representation with Random Distance Matching for Tabular Few-shot Learning (NeurIPS 2024 · similarity 0.86)
- Understanding and Mitigating Memorization in Diffusion Models for Tabular Data (ICML 2025 · similarity 0.86)
- Exponential Family Variational Flow Matching for Tabular Data Generation (ICML 2025 · 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.
