CDMD: A Cross-Dataset Mixed-Type Diffusion Model for Tabular Data

2026-09-30

AI Summary

This study addresses the limitations of existing tabular generative models, which require per-dataset training, exhibit restricted knowledge transfer, and incur high storage costs. To overcome these challenges, this work proposes a cross-dataset mixed-type diffusion model. The method pioneers defining the diffusion process directly within a mixed feature space, dynamically adapting to heterogeneous categorical domains via schema-constrained reverse parameterization. Furthermore, it employs a shared schema-aware Transformer denoiser combined with cascaded diffusion techniques to enable end-to-end joint training. Experimental results demonstrate that the proposed model achieves state-of-the-art average generation quality across seven real-world datasets using fewer parameters. Notably, following large-scale pretraining, the model exhibits significant generalization capabilities on previously unseen datasets, highlighting its potential for universal tabular data synthesis.

Abstract

Generative models for tabular data are typically trained separately for each dataset, limiting knowledge transfer and requiring the storage of many specialized models. In this paper, we introduce CDMD, a tabular diffusion model trained jointly across heterogeneous datasets with different schemas and variable numbers of numerical and categorical features. Unlike existing cross-dataset tabular diffusion models that operate in continuous representation spaces, CDMD defines diffusion directly over the mixed-type feature space and is trained end-to-end. To accommodate heterogeneous categorical domains, we introduce a schema-restricted reverse-process parameterization for masked diffusion models, in which the output space dynamically adapts to each feature's vocabulary. We then compose numerical and categorical feature-level diffusion processes into a schema-dependent row-level process. A shared schema-aware Transformer denoiser captures dependencies between features and parameterizes the reverse process across varying schemas. On seven real-world datasets, a single jointly trained CDMD achieves the highest average generation quality among strong single-dataset and cross-dataset baselines, while using substantially fewer total parameters than the collection of separately trained models. Furthermore, pre-training on a corpus of 337 datasets improves generation on previously unseen datasets under both limited target data and limited adaptation epochs. These results demonstrate the potential of direct mixed-type diffusion for shared and transferable tabular data generation. Our code is available at https://github.com/ketatam/cdmd.

Problem

• Tabular Data
• Generative Models
• Cross-Dataset
• Diffusion Models
• Mixed-Type Features

Innovation

• Cross-Dataset Diffusion Model
• Mixed-Type Tabular Data
• Masked Diffusion
• Schema-Aware Transformer
• Transfer Learning