MIND: Marginal-Invariant Neural Dependency Diffusion for Mixed-Type Tabular Generation
On the map: Diffusion models → Diffusion foundations and sampling also fits: Medical imaging and segmentation
Abstract (arXiv)
This paper proposes MIND, a marginal-invariant neural dependency diffusion model for mixed-type tabular data. MIND does not directly learn the joint distribution in the original heterogeneous feature space. Instead, it first maps different variable types into a unified latent dependency space via column-wise marginal transport. A conditional diffusion model then learns cross-column relationships. Copula-tangent denoising separates known marginal components from learnable dependency residuals. Rank projection during the sampling phase further mitigates marginal shift in reverse diffusion. Experiments across nine diverse tabular benchmarks show that MIND consistently improves marginal fidelity and dependency preservation over existing unified approaches. By explicitly isolating marginal modelling from dependency learning, MIND achieves a strong and stable balance among marginal fidelity, joint dependency preservation, and downstream prediction utility. This work supports separating marginal and dependency modelling as a principled and highly effective paradigm for complex mixed-type tabular generation.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 20 of its 41 references that are in the map. First posted on arXiv 30 Sep 2026.
Builds on
20 of 41 references.
Closest ancestors · 2
Same direction, newest first
- 2021Improved Denoising Diffusion Probabilistic Models18 Feb
- 2020Denoising Diffusion Probabilistic Models19 Jun
Other references · 18
- 2017Attention Is All You Needcited by 59,932
- 2019Modeling Tabular data using Conditional GANcited by 695
- 2022TabDDPM: Modelling Tabular Data with Diffusion Modelscited by 373
- 2017Real-valued (Medical) Time Series Generation with Recurrent Conditional GANscited by 329
- 2022Language Models are Realistic Tabular Data Generatorscited by 290
- 2016Revisiting Classifier Two-Sample Testscited by 258
- 2023Mixed-Type Tabular Data Synthesis with Score-based Diffusion in Latent Spacecited by 183
- 2021CTAB-GAN: Effective Table Data Synthesizingcited by 156
- 2023REaLTabFormer: Generating Realistic Relational and Tabular Data using Transformerscited by 93
- 2022STaSy: Score-based Tabular data Synthesiscited by 85
- 2024TabDiff: a Mixed-type Diffusion Model for Tabular Data Generationcited by 60
- 2023TabMT: Generating tabular data with masked transformerscited by 36
6 more
- 2019Copulas as High-Dimensional Generative Models: Vine Copula Autoencoderscited by 20
- 2024Systematic Assessment of Tabular Data Synthesiscited by 17
- 2025A Survey on Deep Learning Approaches for Tabular Data Generation: Utility, Alignment, Fidelity, Privacy, Diversity, and Beyondcited by 13
- 2014High Dimensional Semiparametric Latent Graphical Model for Mixed Datacited by 10
- 2018Learning Vine Copula Models For Synthetic Data Generationcited by 8
- 2024Structured Evaluation of Synthetic Tabular Datacited by 7
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Measuring the Dependency Gap: Diagnosing Inter-Column Fidelity in Tabular Generative Models
- 2026CDMD: A Cross-Dataset Mixed-Type Diffusion Model for Tabular Data
- 2025TabRep: Training Tabular Diffusion Models with a Simple and Effective Continuous Representation
- 2023Continuous Diffusion for Mixed-Type Tabular Data
- 2025CausalDiffTab: Mixed-Type Causal-Aware Diffusion for Tabular Data Generation
- 2024Diffusion-nested Auto-Regressive Synthesis of Heterogeneous Tabular Data
- 2024Self-Supervision Improves Diffusion Models for Tabular Data Imputation
- 2026AugMask: Training Diffusion Models on Incomplete Tabular Data via Stochastic Augmentation and Masking