{"slug": "geometry-aware-tabular-diffusion", "title": "Geometry-Aware Tabular Diffusion", "summary": "Researchers introduced Geometry-Aware Tabular Diffusion (GATD), a method that improves tabular data synthesis by feeding pairwise angles and lengths from column value differences into diffusion denoisers as inputs and auxiliary targets. The MLP-based model achieved state-of-the-art performance on ten datasets while using 3.5x fewer parameters on average, winning 8/10 Shape, 7/10 Trend, and 9/10 downstream utility benchmarks. The findings demonstrate that explicit relational supervision serves as a portable inductive bias for tabular diffusion, with default loss weights also improving performance across GNN and Transformer denoisers.", "body_md": "arXiv:2606.02607v1 Announce Type: new\nAbstract: Tabular synthesis is critical for privacy-preserving sharing and augmentation, yet diffusion models rely on implicit mechanisms to capture inter-column relationships. We introduce Geometry-Aware Tabular Diffusion (GATD), which augments tabular diffusion denoisers with pairwise angles and lengths computed from column value differences and used as inputs and auxiliary targets. Our MLP instantiation achieves state-of-the-art benchmark performance while using 3.5x fewer parameters on average (up to 25x for classification tasks): on ten datasets, it wins 8/10 Shape, 7/10 Trend, and 9/10 downstream utility (F1/RMSE), reducing Shape and Trend error by 27% and 20%. Default loss weights transfer to GNN and Transformer denoisers, improving Shape on 27/30 and Trend on 25/30 architecture-dataset cells. A matched ablation shows supervision (not extra inputs or capacity) drives the gain. This shows explicit relational supervision is a portable inductive bias for tabular diffusion.", "url": "https://wpnews.pro/news/geometry-aware-tabular-diffusion", "canonical_source": "https://arxiv.org/abs/2606.02607", "published_at": "2026-06-03 04:00:00+00:00", "updated_at": "2026-06-03 04:03:25.006668+00:00", "lang": "en", "topics": ["machine-learning", "generative-ai", "neural-networks", "artificial-intelligence", "ai-research"], "entities": ["GATD", "MLP", "GNN", "Transformer"], "alternates": {"html": "https://wpnews.pro/news/geometry-aware-tabular-diffusion", "markdown": "https://wpnews.pro/news/geometry-aware-tabular-diffusion.md", "text": "https://wpnews.pro/news/geometry-aware-tabular-diffusion.txt", "jsonld": "https://wpnews.pro/news/geometry-aware-tabular-diffusion.jsonld"}}