Constrained Tabular Diffusion for Finance

2026-06-26

AI Summary

This work addresses the critical challenge of generating realistic financial tabular data while strictly adhering to regulatory and business hard constraints—a requirement that existing tabular diffusion models fail to guarantee. The paper introduces, for the first time, a training-free feasibility operator applied during the sampling phase of diffusion models, which dynamically enforces hard constraints throughout the reverse diffusion process. This approach ensures 100% constraint satisfaction without any violations, as demonstrated on large-scale financial datasets. By achieving perfect compliance with domain-specific constraints, the method significantly enhances the practical utility and reliability of synthetic data for applications such as simulation, extrapolation, and regulatory-compliant analytics.

Abstract

Generative models in finance face the dual challenge of producing realistic data while satisfying strict regulatory and economic objectives, a requirement that standard tabular diffusion models cannot provide. To address this difficulty, we introduce Constrained Tabular Diffusion for Finance (CTDF), a novel integration of sampling-time feasibility operations with mixed-type tabular diffusion in financial applications. By incorporating a training-free feasibility operator into the reverse-diffusion sampling loop, CTDF enforces hard constraints for applications such as simulation, legal compliance, and extrapolation. Extensive experiments on large-scale financial datasets demonstrate zero constraint violations and improvement in scarce data utility. CTDF establishes a robust method for generating trustworthy and compliant synthetic data, opening new avenues for rigorous generative modeling and analysis in the financial domain.

Problem

• generative models
• financial data
• hard constraints
• regulatory compliance
• tabular diffusion

Innovation

• constrained diffusion
• tabular data generation
• financial compliance
• feasibility operator
• synthetic data